A daily bite-size selection of top business content.

PM edition. Issue number 1337

Latest 10 stories. Click the button for more.

Term: Joint Business Planning (JBP) - FMCG / CPG

"Joint Business Planning (JBP) in the FMCG/CPG industry is a collaborative, strategic framework where manufacturers and retailers co-create commercial plans, moving away from transactional negotiations to align on shared growth goals. Both parties co-invest in optimized assortments, promotional calendars, and supply chain efficiencies." - Joint Business Planning (JBP) - FMCG / CPG

Pricing disputes, last-minute promotional changes, out-of-stocks, and margin pressure are symptoms of a deeper misalignment between consumer goods manufacturers and retailers. When each side optimises only its own profit and loss, the result is duplicated effort, wasted trade spend, and a fragile relationship that cracks under volatility. Joint Business Planning emerged in fast-moving consumer goods as a way to replace this adversarial pattern with a structured process that links investments on both sides to a single, quantified growth agenda.

From trading terms to shared growth engines

Historically, annual negotiations in FMCG revolved around trading terms: base price, discounts, listing fees, and conditional rebates. Manufacturers committed trade spend in exchange for distribution, shelf space, and loosely defined promotional support. The discussion focused on "how much" rather than "to what end". As promotional intensity rose, both parties found they were spending more but not necessarily unlocking incremental demand or sustainable category growth.

Joint Business Planning (JBP) reorders this sequence. Instead of starting from cost-to-serve and haggling over funding, the conversation begins with category, shopper, and channel objectives and works backwards to the investments, mechanics, and operating model required to deliver them. Trading terms become an enabler of a plan, not the plan itself. This shift is particularly important in FMCG/CPG, where thin margins, high promotional reliance, and retailer consolidation magnify the consequences of misaligned decisions.

In practice, a JBP between a consumer goods company and a retailer is documented in a shared plan that covers business objectives, channel and category roles, assortment and space, pricing and promotion, joint marketing, supply chain and service levels, data and insight sharing, and governance. It typically spans a 12-month horizon, with quarterly reviews and rolling updates, making it a living roadmap instead of a once-a-year spreadsheet exchange.

Substantive definition and scope

Substantively, a Joint Business Plan in FMCG is a structured collaboration process in which a manufacturer and a retailer:

- Define common commercial and category objectives in volume, value, penetration, and share terms.

- Agree a set of initiatives and investments to achieve those objectives, including assortment, merchandising, promotions, pricing, and shopper marketing.

- Clarify roles, responsibilities, and timelines across both organisations, from commercial to supply chain to marketing.

- Establish shared metrics and scorecards to track performance and trigger course corrections.

- Codify these agreements in a single plan that is co-created, signed off, and revisited through formal governance.

Where a traditional trade agreement focuses on "terms and conditions", JBP adds a forward-looking, integrated commercial and operational plan. It is less about who pays for a specific promotion and more about which combination of initiatives yields the highest total return for the category and the joint business.

Practical meaning at the account and category level

For account teams, JBP translates into a structured annual and in-year workflow rather than a single negotiation moment. The typical sequence includes internal alignment, joint discovery, plan design, contracting, and ongoing execution management.

Internal alignment requires the manufacturer to clarify its own brand, channel, and customer strategies before any JBP discussions. This includes understanding the retailer's role in the portfolio (e.g. top 3 growth engine vs marginal account), recent performance, and constraints on trade investment. Retailers undergo a similar process, defining category growth strategies, private-label objectives, and customer experience priorities.

During joint discovery, both sides share views on the "state of the brand" and the "state of the shopper". Manufacturers bring brand equity, innovation pipeline, and media plans; retailers provide performance benchmarks versus peers, shopper demographics, missions, and loyalty or e-commerce data. The goal is not to sell a preset programme but to find overlapping opportunities: under-penetrated shopper segments, emerging subcategories, channel migration, and unmet missions such as quick top-up or health-driven baskets.

Plan design translates these opportunities into specific commercial levers:

- Assortment and space: which SKUs to list, which to rationalise, and how to allocate shelf and digital space by store cluster, with an explicit category role (traffic builder, margin driver, image shaper).

- Promotional strategy: the balance between depth and frequency, mechanics (price cuts, multibuy, bundles), and the alignment with key retailer events and seasonal peaks.

- Pricing and everyday value: a strategy that respects brand positioning, competitive benchmarks, and the retailer's price image, including relative roles for EDLP-style pricing versus high-low tactics.

- Joint marketing: campaigns that combine brand assets with retailer channels (in-store media, CRM, apps, digital banners) and are tailored to shopper missions.

- Supply chain and service: aligned forecasts, minimum service levels, collaborative planning and replenishment, and targeted investments such as shelf-ready packaging or dedicated logistics capacity.

The resulting plan is not just a list of activities but a sequenced calendar linking initiatives to expected commercial outcomes and financial commitments. Execution then becomes a matter of managing variance versus that plan: tracking events, performance, and compliance, and adjusting the programme when shopper behaviour or competitors change.

Core metrics and an analytical backbone

A critical difference between modern JBP and earlier, more relationship-driven collaboration models is the analytical backbone built on data sharing. Retailers and manufacturers increasingly rely on daily or weekly sell-out data, loyalty or panel information, and digital metrics to quantify the impact of their joint actions.

At minimum, a JBP will define target and actual values for key performance indicators such as sales volume, net revenue, gross margin, category share, promotional uplift, and supply chain service levels. More advanced plans incorporate shopper KPIs (penetration, frequency, basket size), media metrics (reach, cost per click, conversion rate), and digital shelf indicators (content quality, availability, search rank).

Analytically, the relationship between investment and outcome is often framed through return metrics. For example, return on marketing investment can be expressed as , where is revenue attributable to a specific joint campaign and is the associated marketing spend. If a joint promotion generates and , then , corresponding to a 300 % return. This kind of calculation allows both sides to compare alternative uses of scarce trade or media budget and to focus the JBP on activities with demonstrable incremental impact.

Similarly, category growth objectives can be framed in terms of baseline and incremental sales. If denotes expected baseline sales without the plan and the incremental volume attributable to JBP initiatives, total planned sales are . The JBP process is, in effect, an attempt to decompose into contributions from assortment, price-pack architecture, promotions, and media, so that investment can be allocated to the most productive levers.

Key parameters and design choices

Designing a robust JBP requires a series of parameter choices, many of which involve trade-offs that reflect each party's risk appetite and strategic priorities.

Time horizon is one such parameter. Most JBPs cover a 12-month period, but some strategic accounts in FMCG operate with rolling, multiyear frameworks to support large investments such as category redesigns or new supply chain infrastructure. A longer horizon enables bolder moves but increases exposure to market volatility and leadership changes.

Investment intensity is another. Manufacturers must decide what proportion of their total trade and shopper budget to allocate to a given retailer, based on that retailer's role in the portfolio, growth trajectory, and relative returns. Retailers judge how much space, promotional support, and data access to provide, balancing one supplier's requests against others and against private-label ambitions.

The degree of data openness sets the ceiling on analytical sophistication. A retailer willing to share near real-time, SKU-level, and shopper-level data enables much richer evaluation of promotional mechanics, cross-category effects, and long-term brand health. Manufacturers, in turn, may share modelling approaches, demand forecasts, and media planning to synchronise investment. However, competitive and privacy considerations mean that JBP rarely involves fully open books; instead, carefully scoped data exchanges and joint analytics are negotiated as part of the plan.

Finally, governance structure determines whether the plan remains active. Leading JBPs define joint steering committees, escalation paths, and regular performance reviews, often monthly for operational KPIs and quarterly for strategic course corrections. Without this structure, the plan risks becoming a static document disconnected from in-store and online execution.

Schools of thought: category-led, retailer-led, manufacturer-led, and data-led

Different organisations and markets have developed distinct philosophies about what JBP should optimise.

A category-led school, influenced by category management, emphasises the category's role as the unit of value creation. The objective is to grow the total category profit pool at the retailer, through better assortment, improved space allocation, and promotions that shift shoppers into higher-value segments rather than simply rotating volume between brands. JBP, in this view, is an extension of joint category planning in which brands accept some cannibalisation in exchange for a larger and more stable category.

A retailer-led school treats JBP primarily as a vehicle to execute the retailer's strategy more efficiently and with higher funding. Here, the retailer defines clear asks around investment levels, support for strategic initiatives (e.g. omnichannel integration, sustainability, private label differentiation), and compliance with standardised processes and scorecards. Manufacturers gain access to better visibility and placement if they align closely with these asks, but they have limited influence over the strategic direction.

A manufacturer-led school, more common where brands hold significant equity or innovation power, uses JBP as a way to secure privileged support for hero SKUs, new platforms, or shopper programmes. The plan is designed around brand growth ambitions, and the retailer is asked to align in return for higher investment, exclusive launches, or differentiated in-store experiences.

The emerging data-led school cuts across these perspectives. Its advocates argue that JBP should start not from power dynamics or broad strategy statements but from granular, shared evidence about what drives incremental value for shoppers and for the joint P&L. Under this approach, hypotheses about assortment, price, and media are tested through controlled pilots; results feed into the next planning cycle, creating a learning loop. This requires both parties to invest in data infrastructure, analytics, and experimentation capabilities.

Tensions and recurring debates

While the rhetoric of partnership is widespread, the practice of JBP in FMCG is shaped by structural tensions that are not easily resolved.

One persistent debate concerns value capture. Even when a plan clearly grows category revenue, the distribution of profit between retailer and manufacturer can be contested. Retailers may view trade funds as a cost of entry and push for higher contributions without guaranteeing equivalent incremental value; manufacturers worry about subsidising baseline sales and eroding brand equity. JBP attempts to solve this through transparent objectives and post-event analysis, but power imbalances and differing internal incentives can blunt its effectiveness.

Another tension lies between standardisation and tailoring. Retail chains seek standardised formats, promotions, and scorecards to manage complexity across hundreds or thousands of stores and dozens of suppliers. Manufacturers, however, want retailer-specific, and sometimes store-cluster-specific, strategies to exploit differential strengths. JBP structures often compromise: a common planning framework and calendar, within which key elements such as assortment, mechanics, and marketing assets are tailored to the retailer's shopper base.

There is also a cultural debate about how collaborative JBP can really be. Some practitioners criticise "paper partnerships" where one side drafts the plan and the other merely signs, undermining the promise of co-creation. Best-practice guidance emphasises building the plan together from the ground up, sharing risks and assumptions, and being explicit about responsibilities and potential conflicts. Yet time pressure, asymmetric information, and negotiation habits mean many JBPs still resemble sophisticated vendor programmes rather than genuine joint ventures.

Measurement is a further source of friction. Agreement on KPIs does not guarantee agreement on attribution. A retailer might credit uplift to its loyalty campaign, while a manufacturer attributes the same gain to a national media burst. Without robust joint measurement frameworks, including controlled tests and shared models, JBP reviews risk devolving into anecdotal scorekeeping instead of evidence-based optimisation.

Why JBP still matters in FMCG/CPG

Despite these challenges, JBP remains strategically important in FMCG/CPG because it offers one of the few mechanisms through which two interdependent but separate organisations can systematically align their decisions in a volatile, low-margin environment.

First, the category and shopper landscape is fragmenting. Growth pockets are shifting towards smaller, more diverse segments, channels, and missions while digital and omnichannel journeys complicate the path to purchase. No single manufacturer or retailer can fully decode these shifts alone. Joint Business Planning provides a forum to combine shopper data, brand insight, and operational capabilities to respond coherently rather than with fragmented, overlapping initiatives.

Second, investment risk is rising. Media fragmentation, inflation, and promotion fatigue make it harder to predict returns on trade and marketing spend. By structuring hypotheses as part of a shared plan, defining expected outcomes, and agreeing in advance how to measure them, JBP allows both parties to take calculated risks instead of playing safe with repeating last year's activities.

Third, execution complexity is increasing. The same brand may run different pack-price architectures across discounters, supermarkets, and e-commerce; promotions might require synchronisation between on-site media, off-site advertising, and in-store activation. JBP offers a way to orchestrate these elements at account level, ensuring that supply chain, digital shelf, field execution, and media plans are mutually reinforcing rather than working at cross purposes.

Finally, JBP matters for organisational learning. The annual planning cycle, when combined with disciplined post-mortems and mid-year reviews, creates a feedback loop in which both sides refine their understanding of what drives incremental value. Over several cycles, this can transform a transactional relationship into a strategic partnership characterised by higher trust, better data sharing, and more innovative joint propositions.

In that sense, the enduring relevance of Joint Business Planning in FMCG/CPG is less about the document created each year and more about the behaviours it fosters: transparency about objectives, shared accountability for outcomes, and a willingness to treat the category and shopper as the true north, even when commercial pressures pull partners back towards zero-sum negotiation.

Quote: John Waldron - Goldman Sachs

"I still think that the vast majority of the price discovery in these transactions will come from the large institutional pools of capital. And so, at the end of the day, that's really the thing that that's going have to go well in the context of making these deals [such as SpaceX, OpenAI and Anthropic] work." - John Waldron - Goldman Sachs

Control over where and how prices are set in late-stage technology financings has become one of the defining questions for the next decade of markets. It determines who captures upside from artificial intelligence breakthroughs, how resilient capital structures will be when sentiment turns, and whether the path from private valuation to public listing remains orderly or becomes a series of dislocations. In the case of AI champions such as SpaceX, OpenAI, and Anthropic, the sheer scale of capital required, the concentration of economic power, and the opacity of private rounds make the locus of price discovery a critical systemic issue rather than a technical footnote.

The underlying tension is simple: AI platforms with global reach need tens of billions in compute, data, and talent investment, yet their equity prices are being inferred from relatively thin slices of secondary trading and bespoke private transactions, not from broad and continuous public markets. The question is whether retail and smaller pools of speculative capital can meaningfully set prices in this environment, or whether large institutional investors will remain the primary mechanism through which valuations are negotiated, disciplined, and transmitted into the wider market. For financial intermediaries that underwrite, structure, and distribute such deals, the answer determines both risk management and revenue opportunity.

From public markets to private price signals

For most of the twentieth century, price discovery for major corporates was anchored in public exchanges, with day-to-day prices shaped by a mix of institutional flows, retail activity, and specialist dealers. The microstructure was noisy, but there was at least a transparent quote for almost every security at almost every moment. Over the past 20 years, however, late-stage venture and growth equity markets have grown to a scale where the most important technology firms can delay public listings for far longer and can raise 5, 10, or even 20 billion in private capital before an initial public offering. That shift has relocated a substantial part of price discovery from open exchanges to private deal rooms.

In those rooms the dominant actors are not day traders but sovereign wealth funds, large asset managers, pension schemes, insurance companies, and major hedge funds. These institutions write cheques measured in billions, not millions. Their internal models, risk committees, and benchmarking processes determine the ranges at which boards are willing to transact. Even when smaller investors participate via feeder funds or structured vehicles, the indicative price is typically set by a handful of large anchors whose willingness to provide capital at a given valuation is decisive.

In principle, public markets might eventually overwrite these private valuations once a listing occurs. In practice, the initial price discovery at IPO is also heavily shaped by institutional demand. Book-building processes rely on indications of interest from a curated list of long-only funds, hedge funds, and other professional investors. Retail orders, even when meaningful in aggregate, usually follow the range signalled by this group. That means the same large pools of capital that dominated late-stage private rounds often remain central in the first liquid price that the market sees.

AI platforms and the capital intensity of scale

The dynamic is amplified in AI. Training frontier models, building global data centres, and integrating AI into industrial and consumer workflows require extraordinary upfront expenditure. Cloud credits and partnership structures can ease the burden, but for independent or semi-independent entities such as OpenAI or Anthropic, the capital stack must still support long-duration research and infrastructure commitments. SpaceX, while a space and launch company rather than a pure AI play, shares a similar profile: massive capex, long development cycles, and complex regulatory risk.

These companies are therefore engaged in a multi-round capital formation process in which each new raise both funds operations and crystallises an implicit valuation step. When a round prices OpenAI or Anthropic at, say, 60 billion or 80 billion, that number is not a market consensus but a negotiated outcome between management and a relatively small syndicate of investors. It may reference comparable multiples from public AI-exposed companies and discounted cash-flow scenarios for long-run monetisation, but the final figure is fundamentally a deal term. Subsequent secondary trades by employees or early backers tend to cluster near that negotiated level, unless a macro shock or major news event forces a repricing.

Because the needs are so large, the marginal buyer that matters in these rounds is almost always institutional. A single sovereign wealth fund or mega-fund can bridge a financing gap of several billions; thousands of smaller players cannot coordinate quickly enough or accept the same information constraints. That concentration of bargaining power means that price discovery is heavily conditioned by the opportunity cost of capital for these institutions, their internal risk constraints, and their forward view on AI adoption curves. Retail enthusiasm in listed AI-adjacent names may inform sentiment, but it rarely changes the cheque sizes or target returns required by the limited partners who ultimately supply the money.

Who John Waldron is, and why his perspective matters

The speaker of the quote occupies a vantage point that is not merely observational but operational. As president and chief operating officer of a major global investment bank, John Waldron oversees a platform that arranges, underwrites, and distributes some of the world's largest equity and debt transactions. His firm advises both issuers such as SpaceX and AI start-ups, and the institutional investors that buy into these deals. It also operates trading, prime brokerage, and asset management businesses that interface with hedge funds, pension funds, and other large pools of capital.

From this position, Waldron is acutely aware of the difference between headline narratives in media or social platforms and the practical conditions under which multi-billion capital commitments are made. When he emphasises institutional pools of capital as the primary source of price discovery, he is effectively describing the way his own franchise operates: block trades negotiated with a small number of lead accounts; structured IPO allocations built around core anchor investors; and private rounds in which a single cornerstone order can validate or derail a proposed valuation.

His perspective is also informed by risk. As a senior executive, he bears responsibility for firm-wide balance sheet usage, reputational risk, and client outcomes. Deals that clear at artificially inflated prices may generate short-term fees, but they also raise the probability of painful down-rounds, post-IPO collapses, and strained relationships with both issuers and investors. Insisting that institutional price discovery go well is therefore not only a prediction; it is a condition for the long-term sustainability of the franchise and, more broadly, for market stability.

The mechanics of institutional price discovery

In the background, price discovery by large institutions combines hard modelling, peer comparables, and soft information from networks and management interaction. Analysts will typically build cash-flow projections under multiple scenarios for adoption, margin evolution, and regulatory constraints. Even when the business is early-stage and cash-flow negative, they attach probabilities to different outcome trees, discount them at hurdle rates that reflect perceived risk, and derive a range of plausible enterprise values. The conversation then becomes one of where in that range the issuer and the investor can agree.

For AI companies, the modelling challenge is acute. Revenue visibility may be limited, with usage-based pricing and platform economics that depend on partner ecosystems still being formed. Compute costs can change rapidly as hardware generations evolve. Regulatory shifts on data usage, safety requirements, and liability could materially alter future profitability. As a result, institutional investors often use scenario-based frameworks. They might consider a base case in which enterprise adoption follows a gradual S-curve, a bull case in which AI rapidly unlocks productivity across multiple sectors, and a bear case in which regulatory or technical setbacks slow deployment.

Within such frameworks, what matters most is not a single point estimate but the sensitivity of valuation to key assumptions. An investor may accept a higher headline valuation if given structural protections such as liquidation preferences, anti-dilution provisions, or governance rights that allow for intervention if execution falters. Conversely, a governance-light structure might require a lower entry price to compensate for the inability to influence outcomes. These trade-offs are rarely visible to public observers yet are central to the effective price paid for risk.

Strategic and technological tensions in AI financing

Price discovery in AI deals is not a purely financial exercise; it is entangled with strategic alliances, sovereignty concerns, and technological competition. Large cloud providers that partner with AI firms may participate in equity financing at valuations that reflect strategic value rather than traditional investment metrics. Governments and quasi-sovereign funds may consider national competitiveness in AI when deciding whether to support domestic champions. These considerations can push prices up relative to what a purely financial analysis would suggest.

For intermediaries, this creates a tension between facilitating strategic capital and maintaining valuation discipline. If one investor is willing to pay a premium to secure preferential access to models or infrastructure, others must decide whether to follow at that price or to walk away and risk being structurally underexposed to a transformative technology. In hot markets, fear of missing out can erode discipline; in cooler periods, risk aversion can overshoot in the other direction. Waldron's focus on institutional pools as the key site of price discovery implicitly recognises that managing this tension requires actors with both the scale and the governance structures to resist the most extreme swings.

There is also a technological feedback loop. High valuations enable AI companies to raise more capital, hire more researchers, and purchase more compute, reinforcing their lead. However, they also set high expectations for growth and profitability that may be difficult to achieve in practice. If institutional investors collectively overestimate the addressable market or underestimate the regulatory drag, a later correction could be sharp. Conversely, if they are too conservative, they risk slowing down innovation or pushing companies to seek funding from less transparent sources.

Debates and objections: can retail and crypto-style markets do more?

Not everyone accepts the premise that institutional capital should dominate price discovery. The experience of meme stocks, cryptoassets, and retail-driven speculative booms has shown that large swarms of small investors, coordinated through digital platforms, can move prices dramatically. Proponents argue that this democratises finance, surfaces diverse information, and challenges potential conflicts of interest among large intermediaries.

However, the suitability of such mechanisms for financing companies like SpaceX, OpenAI, and Anthropic is contested. These businesses require patient capital willing to tolerate long periods of negative free cash flow, complex technical risk, and binary regulatory outcomes. Retail flows are often momentum-driven and may be less tolerant of prolonged volatility or extended lock-ups. Moreover, the disclosures required for public retail ownership may conflict with the competitive need to keep certain technical and strategic information confidential until a more mature stage.

There are also concerns about systemic risk. If retail-driven price discovery sets valuations at levels far above what institutional fundamentals would justify, the eventual correction could erode trust not only in specific names but in AI as an investable theme. This in turn might complicate financing for more grounded, less glamorous AI applications in industry and public services. Institutional gatekeepers, for all their flaws, provide a buffer between sentiment and capital allocation, absorbing information and adjusting positions in a more measured fashion.

Why the success of institutional price discovery matters

When Waldron says that this mechanism has to go well for the deals to work, he is pointing to several layers of dependency. First, the issuers themselves need valuations that are high enough to fund ambitious plans without excessive dilution, but not so high that every future round becomes an exercise in defending a fragile narrative. Second, investors require returns that compensate them for risk, within mandates that are accountable to pensioners, policyholders, and endowments. Third, intermediaries need both sides to feel fairly treated if they are to maintain long-term client relationships and reputational capital.

If institutional price discovery fails - because models prove systematically over-optimistic, governance proves too weak to correct course, or strategic pressures override financial discipline - a wave of disappointing outcomes could follow. Down-rounds in private markets can demoralise employees whose equity is underwater, prompt litigation, and limit future fund-raising options. Public market disappointments, especially after heavily marketed IPOs of AI leaders, can spark political scrutiny, regulatory tightening, and demands for stricter control over both finance and AI development.

Conversely, if it succeeds, the benefits extend beyond the firms directly involved. A credible valuation structure for AI unicorns provides benchmarks against which smaller companies can be priced, informs M&A activity, and helps public investors calibrate multiples for AI-exposed incumbents. It also gives policymakers clearer signals about where private capital sees sustainable value, which can guide decisions on public research funding, infrastructure investment, and education priorities.

The evolving architecture of capital pools

It is also important to recognise that the category of "large institutional pools of capital" is itself evolving. Traditional pension and insurance funds are being joined by sovereign wealth funds from emerging markets, large family offices with institutional-grade capabilities, and specialist AI or technology funds that raise many billions from global limited partners. Some of these actors operate with longer horizons and higher risk tolerance than classic institutions; others are more constrained by domestic politics or public scrutiny.

The result is a more heterogeneous set of views feeding into price discovery. For AI deals, this diversity can be helpful, as it allows issuers and intermediaries to match specific risk-return propositions with compatible investors. Yet it also means that coordination is harder. A single change in regulatory perception or geopolitical alignment can reprice risk for a subset of investors, shifting demand and creating cross-currents in valuations. Managing such complexity is part of why major investment banks emphasise their role as translators between issuer needs and investor constraints.

In this context, Waldron's focus on these pools is both descriptive and aspirational. Descriptive, because they are already the primary providers of bulk capital for AI and space-related deals. Aspirational, because the hope is that their processes, governance, and experience will be robust enough to steer valuations through cycles of hype and disappointment, ensuring that transformative technologies are funded at scales commensurate with their potential without triggering destabilising bubbles.

Implications for future IPOs of SpaceX, OpenAI, and Anthropic

When considering eventual IPOs or other liquidity events for SpaceX, OpenAI, and Anthropic, the present configuration of private price discovery sets the starting conditions. If late-stage rounds have been anchored by rigorous institutional assessment, public market investors may see the transition as a natural extension of existing valuation logic. If, instead, valuations have floated far ahead of fundamentals, the listing event may become a de-rating rather than a celebration.

For SpaceX, questions about launch cadence, satellite broadband economics, and defence and intelligence contracts will shape institutional models. For OpenAI and Anthropic, topics such as model commoditisation, cloud revenue-sharing arrangements, and the balance between safety obligations and monetisation will dominate. In each case, the terms on which institutions are willing to finance the next phase of growth - pre-IPO and at IPO - will reflect how they resolve these uncertainties.

The broader market will watch these processes closely, not only because of the companies' intrinsic importance but because their deals will serve as reference points for a whole generation of AI-related offerings. If the transactions clear smoothly, with robust books and post-listing performance that broadly matches expectations, the template will reinforce the central role of institutional capital in orchestrating innovation financing. If they misfire, pressure will grow for alternative mechanisms of price discovery, whether through more retail participation, more direct state involvement, or new forms of digital asset-based funding.

In that sense, the quality of institutional price discovery for these AI and space leaders is not just a technical matter for bankers and portfolio managers. It is one of the levers through which societies decide how quickly to scale transformative technologies, how to distribute the economic gains they generate, and how to allocate the risks that accompany them. The stakes extend far beyond any single deal.

Term: Jobs to Be Done (JTBD) framework - FMCG / CPG

"In FMCG/CPG, the 'Jobs to Be Done' (JTBD) framework posits that consumers do not simply buy products, but rather 'hire' them to make progress or solve a specific problem in a given circumstance. By shifting focus away from traditional customer demographics and toward the functional, emotional, and social motivations behind a purchase, this consumer-centric philosophy uncovers unspoken needs." - Jobs to Be Done (JTBD) framework - FMCG / CPG

Most FMCG categories suffer from crowded shelves, marginal differentiation, and relentless price pressure. Yet some brands cut through, commanding loyalty and price premia despite offering near-identical features. The underlying driver is rarely a clever flavour extension or an eye-catching pack; it is whether the product helps people reliably achieve the progress they are seeking in a specific situation. That progress may be small and mundane, like getting out of the door faster on a weekday morning, or more identity-laden, like feeling like a competent, caring parent. When teams fail to decode these underlying goals, they mistake surface behaviours for true demand, and innovation becomes guesswork.

From categories and consumers to situations and progress

Conventional FMCG strategy starts with consumer segments and product categories: young urban families in the breakfast cereal market; health-conscious millennials in beverages; value-seeking shoppers in household cleaning. Demographics and attitudinal segments remain useful, but they often say little about what triggers an actual purchase at a specific moment or why people switch between options that sit in different categories but compete for the same use case. A rushed commuter may grab a chocolate bar, a protein shake, or a pastry from the same shelf because the underlying objective is to stay energised and avoid hunger until lunch, not to consume a particular category.

The Jobs to Be Done perspective reorganises this picture around circumstances and progress. Instead of asking "Who is our consumer?" brand teams examine recurring situations in which people feel a gap between where they are and where they want to be. The gap might concern a functional outcome (clean a kitchen quickly), an emotional state (feel proud when guests arrive), or a social perception (avoid seeming careless or unhygienic to others). Products become tools people "hire" to bridge that gap; rival products, and even non-consumption, become alternative ways of getting the job done.

This shift has several practical implications for FMCG and CPG businesses. Category boundaries become less important than competitive sets defined by shared jobs. Research designs evolve from measuring likes and preferences to mapping decision contexts and constraints. Innovation briefs move away from "develop a premium variant" and toward "help parents reduce weekday breakfast stress within 5 minutes and without extra washing up". Marketing stories pivot from describing product features to narrating how the brand fits into real-life scenarios of progress.

The substance of a "job" in FMCG

A job is not a vague desire for a product, nor a loyalty to a brand, but a description of the progress someone wants in a given situation without assuming any specific solution. In the FMCG context, jobs typically combine three intertwined dimensions:

- Functional: what concrete outcome the person needs, such as removing stains, satisfying hunger, freshening breath, or soothing a child's skin.

- Emotional: how they want to feel during and after the event, for example reassured, in control, pampered, or less guilty.

- Social: how they wish to be perceived by others, such as being seen as a good host, a responsible parent, or a savvy, eco-conscious shopper.

Consider a laundry detergent bought before a family gathering. The core functional job may be "ensure clothes look and smell clean", but there is also an emotional component (avoid the anxiety of being judged) and a social dimension (signal that the household is well kept). A budget detergent that excels only on functional cleanliness may underperform if it fails to support the emotional and social aspects of the job as experienced in that circumstance.

Crucially, the same consumer can "hire" very different products for distinct jobs across the week. A shopper might buy a budget bulk coffee for weekday mornings (job: "wake up sufficiently and save money") and a premium single-origin pack for weekends (job: "create a small ritual of pleasure and self-treating"). Demographic data remains identical, but the job, context, and evaluation criteria change significantly.

Decomposing jobs: outcomes, steps, and metrics

One influential stream of JTBD thinking distinguishes between the job as a process and the outcomes consumers use to assess success. In this view, a core functional task can be broken into sequential steps that describe how people try to get it done, and each step has associated desired outcomes that act as metrics.

For example, the job "get a quick, satisfying weekday breakfast" can be mapped into steps such as planning, shopping, storing, preparing, consuming, and cleaning up. Each step involves specific success criteria: minimise time to prepare, reduce mess, avoid hunger before lunchtime, avoid complaints from children, limit sugar intake, and so on. From a JTBD standpoint, these desired outcomes effectively define the spec against which consumers judge competing solutions.

Although day-to-day FMCG work rarely requires heavy formal mathematics, this outcomes view can be expressed conceptually. Let denote a particular job and denote the set of desired outcomes associated with that job. For a given product , a consumer intuitively evaluates performance against each outcome, yielding perceived scores for . The overall suitability of for job can be thought of as a function , where is the perceived utility of "hiring" that product for the job.

Different shoppers and situations weight these outcome-scores differently. A busy parent might care more about "time to prepare" and "mess", while a fitness-focused individual emphasises "satiety" and "nutritional profile". Explicitly articulating outcomes and their relative importance allows CPG teams to see where existing offerings overserve on some dimensions and underserve on others, creating space for repositioning or new propositions.

Parameters that define a job in FMCG practice

In applied CPG work, several parameters usually define and differentiate jobs:

- Situation: when and where the job arises (weekday breakfast at home, late-night snacking on the sofa, cleaning just before guests arrive, on-the-go hydration during commuting).

- Trigger: what event or feeling initiates the job, such as hunger, embarrassment about odours, worry about germs, time pressure, or boredom.

- Constraints: budget limits, time windows, dietary restrictions, household equipment, and store availability that shape what is realistically hireable.

- Desired outcomes: the specific functional, emotional, and social end states by which success is judged.

- Existing alternatives: not just other brands in the same category, but cross-category solutions and workarounds, including non-consumption (skipping breakfast, using water instead of a specialised cleaner).

Capturing these parameters requires methods that observe or reconstruct behaviour in context: in-home interviews, shop-alongs, usage diaries, and ethnographies that probe not only what is purchased but what was considered, rejected, or never even noticed. Quantitative research can then size how common particular jobs and outcomes are, estimate their frequency and economic value, and link them to observed purchasing behaviours.

Major schools of thought within JTBD

Within the broader Jobs to Be Done landscape, several schools of thought influence how FMCG teams apply the concept.

One stream, often associated with Tony Ulwick, emphasises jobs as stable functional tasks and uses a structured needs framework. It distinguishes types of customers (job executors, those supporting the product lifecycle, and buyers) and breaks jobs into steps to surface measurable desired outcomes. This approach tends to lead to detailed job maps and quantitative opportunity scoring, where teams systematically assess which outcomes are underserved or overserved and prioritise innovation accordingly.

Another stream, linked to Clayton Christensen, puts more weight on the narrative of hiring and firing solutions. It focuses on switching moments, the stories of when people stop using one solution and adopt another, and the emotional and social frictions that drive this change. In this view, understanding the "job" is closely tied to understanding the circumstances under which people re-evaluate their options and reconfigure their routines.

A third, more design-oriented strand emphasises qualitative discovery in context. It treats JTBD as an empathy-building lens embedded into design research, experience mapping, and service blueprints. Consumer researchers and UX teams use job narratives to identify friction points in the whole experience of buying, using, and disposing of FMCG products, not just in the moment of consumption.

These perspectives are complementary rather than mutually exclusive. CPG organisations often blend them, using qualitative hiring stories to define jobs and then applying a structured outcomes framework to quantify the opportunity for new products, pack formats, or claims.

Debates, tensions, and misapplications

Because the JTBD label has become popular, practice varies widely and has generated several debates relevant to FMCG.

Job definition scope. One tension concerns how broad or narrow a job should be. Defining a job too narrowly ("open this particular type of pouch") can reduce it to a feature-level task and obscure more strategic insights. Defining it too broadly ("live a healthier life") yields aspirations too diffuse to guide specific innovation. In CPG, a useful middle ground tends to be the recurring scenario that is stable enough to design for but specific enough to expose concrete trade-offs, such as "provide a quick after-school snack that feels like a treat but is not seen as unhealthy".

Functional bias vs emotional and social layers. Another debate arises around the balance between functional and non-functional elements. Some interpretations foreground functional steps and metrics, risking underestimation of emotional and identity-driven motivations, which are often decisive in personal care, baby care, and indulgent food categories. Others focus so much on emotional storytelling that they lose the operational precision needed to design packaging, formula, or merchandising that reliably changes behaviour. Robust FMCG work tends to insist that every job description explicitly include all three dimensions.

Jobs vs segments. JTBD thinking sometimes invites the claim that traditional segmentation should be discarded. In practice, FMCG teams usually need both. Jobs describe situations and progress; segments describe clusters of people who are more likely to experience particular jobs, hold certain outcome priorities, and accept specific constraints. Ignoring segments can create propositions that are conceptually strong but economically weak if the job occurs rarely within high-value consumers. Ignoring jobs leads to segments that are easy to describe yet hard to activate because they do not map to moments of decision in store or online.

Over-simplification. A further risk is treating the "hire" metaphor literally and looking for a single dominant job per product, when many FMCG items serve multiple jobs across occasions. A multi-pack of snack bars may be used for children's lunchboxes, adult office snacks, and emergency on-the-go meals, each with different outcome priorities. Forcing a one-job narrative can blind teams to profitable secondary jobs or to tensions between them that require separate variants or pack designs.

Why the framework still matters in FMCG and CPG

Despite waves of new methodologies and data sources, JTBD remains influential because it tackles a persistent gap: the difficulty of linking observed behaviour to underlying motivation in a way that is actionable for product, packaging, and marketing decisions.

First, it improves innovation hit rates by grounding ideas in recurring jobs rather than abstract category trends or technology push. When teams start from a carefully researched job, they can articulate design specifications in terms of outcomes: for example, reducing the variance of preparation time, limiting the number of steps, or increasing perceived control in mess-prone tasks. Even without formal equations, thinking in terms of outcomes and constraints forces clarity on what a new product must improve and what trade-offs are acceptable.

Second, it reframes competition. From a jobs lens, a ready meal competes not only with other ready meals but with takeaway, meal kits, and the decision to skip cooking entirely. A fabric freshener may compete with full washing cycles and with changing clothes. CPG players that understand this can identify under-served jobs where their categories barely feature today and design offerings that insert the brand into those occasions.

Third, it enhances communication and branding. Marketing built around jobs answers the question "when and why should I think of this product?" by showing recognisable situations and the progress achieved, instead of listing generic features. This orientation aligns with the view that effective marketing tells the story of what the product does for the consumer and how it fits into their life, not just what it contains or how it is made.

Fourth, it provides a shared language across functions. R&D, insights, brand, design, and sales teams can align more easily around well-defined jobs than around abstract brand values. A statement such as "we are targeting the job of helping young professionals assemble a healthy weekday lunch in under 8 minutes with minimal cognitive effort" is clearer for packaging designers, product developers, and trade marketing than a broad aspiration like "own health-conscious convenience".

Incorporating JTBD into FMCG research and decision-making

Embedding JTBD thinking in CPG organisations usually involves layering it onto existing research and development processes rather than replacing them.

- Qualitative discovery. Teams use in-home or digital diaries, accompanied shops, and immersion sessions to uncover recurring jobs, triggers, and workarounds. Interviewers focus on specific episodes: "Tell me about the last time you..." rather than abstract preferences, and probe what other solutions were considered or tried.

- Job definition and mapping. Insights and cross-functional teams synthesise narratives into a concise job statement, a stepwise job map, and a list of desired outcomes. They explicitly separate functional, emotional, and social components to avoid collapsing them into a single label.

- Quantitative sizing. Surveys or behavioural data are used to estimate how many people experience each job, how often, what they currently hire, and which outcomes are most important yet least satisfied. Opportunity scores can be computed by combining prevalence, importance, and dissatisfaction measures, helping prioritise where to innovate first.

- Concept development and testing. New product ideas, pack formats, or claims are evaluated against the defined job and outcomes: does the concept improve key outcomes without worsening others beyond acceptable trade-offs? Concept tests can explicitly frame scenarios that match the job to ensure respondents evaluate the proposition in the right context.

- Ongoing refinement. As products launch, consumption data, reviews, and further qualitative feedback are used to update the job understanding. Shifts in culture, technology, and retail environments may create adjacent jobs or change constraints, prompting iteration.

Done consistently, this approach cultivates a portfolio view of jobs rather than just of categories, allowing CPG businesses to track where they are strong or weak across the spectrum of everyday situations in which consumers seek help.

Enduring relevance in a changing marketplace

Digital grocery, direct-to-consumer brands, and subscription models have expanded how people access FMCG products, but they have not changed the basic reality that purchase decisions are rooted in attempts to make discrete progress in specific circumstances. If anything, the explosion of choice and information makes it more important to anchor innovation and marketing in clearly understood jobs. Algorithms can optimise assortments and promotions, but they still require a human understanding of what problems products solve and for whom.

By focusing on what people are genuinely trying to accomplish, how they evaluate success, and what stands in their way, the Jobs to Be Done lens continues to offer FMCG and CPG teams a disciplined way to cut through noise. It redirects attention from demographic stereotypes and feature lists to lived situations, trade-offs, and progress, keeping consumer packaged goods grounded in the everyday lives they are meant to serve.

Quote: John Waldron - Goldman Sachs

"If there are problems in software, it will first start to show up in the riskiest part of the capital structure, which is equity. So that stands to reason. To me, too much press has been spilled on private credit and not enough focus on the fact that there's a whole bunch of equity underneath that that is the first loss position, and that's where you probably want to pay the most attention earliest." - John Waldron - Goldman Sachs

Stress in software and technology financing rarely appears first where commentary is loudest. It tends to surface where contractual protection is weakest, cash flow sensitivity is highest, and valuations have drifted furthest from fundamentals. In contemporary capital structures supporting sponsor-backed software companies, that weak link is typically common equity, not the private credit sitting above it . Understanding why requires tracing how software businesses were capitalised through the low-rate era, how private credit stepped into the vacuum left by banks, and why equity investors quietly became shock absorbers for the entire structure.

The factual backdrop is a decade in which recurring-revenue software became the poster child for aggressive leverage. Ultra-low policy rates, abundant venture and growth equity, and a wave of sponsor-led buyouts combined to push valuation multiples to levels that assumed durable double-digit growth and minimal cyclicality. Private credit managers rushed to fund these deals, confident that high gross margins and subscription models would protect cash flows, while covenant packages were progressively loosened. Recent research from Goldman Sachs notes that roughly 23 % of the private credit market is now tied to software loans, a concentration that even industry insiders characterise as excessive by historical standards . Against that backdrop, the remark that problems will show up in equity first is not a reassurance so much as a reminder of how the hierarchy of pain actually works.

The speaker, John Waldron of Goldman Sachs, sits at the junction of several relevant vantage points: advisory to private equity sponsors, underwriting and distribution of leveraged finance, and a rapidly expanding private credit platform that now manages roughly USD 145 billion in private credit exposure . The firm has reorganised its global banking and markets unit into a Capital Solutions Group designed explicitly to intermediate across financing, origination, structuring, and risk management for corporate and sponsor clients . That reorganisation underscores an institutional bet that privately negotiated credit and hybrid financing will remain central to the funding of software and other growth sectors, even as public markets remain volatile.

The substantive claim embedded in the remark is straightforward: equity is structurally the first-loss tranche in a typical leveraged capital structure. Debt instruments in private credit deals are generally senior secured, with priority claims on assets and cash flows, contractual interest payments, and in many cases maintenance covenants that give lenders the right to intervene early. Equity, by contrast, has no contractual claim to cash, absorbs valuation volatility, and can be diluted, written down or wiped out without triggering defaults on the debt. When a software company encounters operational or growth problems, the first place the stress is visible is equity pricing, sponsor mark-downs, and downward revisions of forward projections, long before a payment default appears in the loan book.

To see this more formally in a finance framework, one can model the equity of a levered firm as a call option on the enterprise value. Let denote the enterprise value at time , the face value of debt maturing at , and the risk-free rate. In a structural model, equity can be represented as:

where is the cumulative normal distribution and:

Here is the volatility of enterprise value. This framing makes Waldron's intuition transparent. As rises or expectations for fall due to weaker software fundamentals, equity values react sharply because they are effectively out-of-the-money or near-the-money options. The debt value, especially when senior secured, is comparatively insensitive until the probability that becomes meaningful. In other words, early tremors show up in the equity option; the credit claim is hit later.

Software business models appeared to offer unusually favourable parameters for this option-like equity: high gross margins, recurring revenue visibility, and scalable cost structures. Sponsors extrapolated these characteristics into aggressive capital structures, assuming that revenue growth and pricing power would sustain interest coverage and allow rapid de-leveraging. Private credit funds, competing to provide unitranche and first-lien loans, often underwrote based on forward-looking pro forma EBITDA and contractual renewals rather than historic downturn performance. As long as software multiples remained high, equity cushions looked deep and reassuring.

However, that apparent cushion is acutely sensitive to small shifts in top-line growth and discount rates. Consider a simplified discounted cash flow representation for a software firm where enterprise value is approximated by:

with representing next period free cash flow, the perpetual growth rate and the discount rate. For high-growth software companies priced as if is only marginally below , even modest reductions in or increases in can trigger sharp contractions in . If the firm carries substantial debt , the equity value can be compressed or eradicated long before the debt principal appears at risk. That is the mechanical reason why equity is where deterioration manifests first.

The post-pandemic tightening cycle intensified this sensitivity. Rising base rates increased funding costs, while inflation and wage pressures squeezed operating margins. At the same time, growth expectations for many software companies were revised downward as customers rationalised licences and delayed digital transformation projects. The equity markets reacted quickly: multiples for many high-growth names compressed, IPO windows partially shut, and late-stage venture funding became more selective. Yet the private credit portfolios financing the same companies often looked stable on the surface because borrowers continued to make interest payments and because portfolio valuations lagged public market repricing.

Waldron's criticism that commentary has focused too heavily on private credit and too little on the equity beneath reflects this asymmetry in information and narrative. Private credit is relatively opaque: positions are not traded with public price discovery, covenants are confidential, and managers disclose performance with a delay. This opacity invites concern that a hidden bubble may be building among software loans. But in many leveraged software platforms, the true point of fragility is not the senior secured loan paying a 7 % to 9 % cash coupon; it is the equity slug that funded the acquisition at a 20x revenue multiple based on aggressive growth plans. When those plans falter, equity valuations must be recalibrated, and sponsors may find themselves injecting additional capital, selling assets, or accepting marked-down exits.

The creation of Goldman Sachs's Capital Solutions Group is a strategic response to exactly this environment . By combining financing, origination, and risk management, the firm is effectively positioning itself to manage the full stack of capital from senior loans through mezzanine to equity and preferred instruments. That multi-layered vantage point reveals where risk truly sits in a structure. For a typical sponsor-backed software company, the capital stack might include a senior secured term loan, a revolving credit facility, possibly a second-lien or mezzanine tranche, and a substantial equity contribution from the sponsor and management. The private credit piece grabs headlines because of its size and growth, but the equity is what first absorbs the impact when revenue growth slows, churn rises, or new competitors erode pricing.

The technological context also matters. Software lending has been buoyed by the perceived defensiveness of subscription revenue and the rise of software-as-a-service in mission-critical functions. Yet the AI wave has created new uncertainty. On the one hand, incumbents may face disruption as AI-native competitors undercut them or offer better features. On the other, AI-driven productivity tools promise efficiency gains, but also require capital expenditure and strategic repositioning. Goldman's own commentary on AI investment expectations suggests that it anticipates continued robust capital expenditure on AI infrastructure, particularly led by US hyperscalers . For software firms dependent on those platforms, changes in pricing, infrastructure costs, or competitive dynamics can rapidly alter unit economics, again feeding through to equity valuations long before credit metrics breach covenants.

Debate around private credit risk often centres on whether underwriting standards have weakened and whether concentration in sectors like software is dangerous. Critics argue that covenant-lite structures, aggressive leverage multiples, and a lack of mark-to-market transparency could store up trouble. Proponents counter that private credit managers have tighter relationships with borrowers, better information rights, and the ability to work constructively through periods of stress. Waldron's emphasis subtly reframes the debate: even if some stress does emerge in software-linked private credit, it is unlikely to be the first or primary place where losses accumulate. Instead, equity investors in highly levered software deals are already experiencing a reset that will change the risk profile of the debt above them.

From a risk-transfer perspective, this is simply the pecking order of claims operating as designed. In a stylised capital structure, suppose a software company has enterprise value , senior debt , junior debt , and equity . If operational issues reduce to , the entire decline hits equity first, leaving while both tranches of debt remain fully covered. Only once do junior lenders face principal impairment, and senior lenders are not at risk until falls below . Empirically, that means that by the time private credit portfolios start to show meaningful realised losses, equity investors are likely to have endured a prolonged period of write-downs, down-rounds, and unfavourable exits.

One objection to the argument is that equity valuations in private markets are themselves opaque, so relying on equity as an early warning signal may not be straightforward. Sponsor marks can lag reality, especially where there is no transaction to force a repricing. However, several practical indicators exist: reduced bidding for new software assets, an increase in broken auction processes, widening gaps between buyer and seller expectations, more frequent use of structured equity solutions, and sponsors injecting additional preferred equity to support over-levered portfolio companies. These behaviours are the qualitative signs that equity is absorbing stress. Lenders involved in capital solutions work will see these signals well before a payment is missed on a private credit instrument, which strengthens Waldron's case that attention should be directed there.

Furthermore, software as a sector amplifies equity risk because its intangible asset base offers limited hard collateral. When lenders underwrite against recurring cash flows and customer contracts, recovery values in a default scenario are highly uncertain. Source code, customer lists, and data have value, but often far less than the optimistic projections baked into the original deal. This makes the presence of a deep equity cushion more critical: equity holders are the ones effectively underwriting the uncertainty in those intangible values. A thin or rapidly eroding equity layer should therefore be a red flag not only for sponsors but for lenders who have relied on that buffer as part of their loss-absorbing structure.

The comment also points to a broader narrative tension: the media appeal of a looming private credit crisis versus the less dramatic but more probable story of prolonged equity pain and restructuring in software portfolios. Predicting a systemic private credit event is headline-grabbing, yet the more subtle reality may be a drawn-out process of sponsor-led recapitalisations, minority stake sales, and operational turnarounds that gradually reconcile inflated entry valuations with more modest cash flow outcomes. For institutions like Goldman Sachs, which straddle advisory, lending, and asset management, the objective is not only to avoid losses but to position themselves as indispensable intermediaries in this adjustment process.

In that sense, the expansion of Goldman's private credit and alternatives capabilities is both a risk and an opportunity . On the one hand, increased exposure to software-heavy private credit portfolios means the firm is deeply intertwined with the sector's fortunes. On the other, the ability to deploy fresh capital into recapitalisations, structure preferred equity or convertible instruments, and engineer liability management transactions allows it to influence where value ultimately settles in the capital structure. If equity investors have already absorbed substantial losses, new capital providers can negotiate favourable terms, effectively resetting the stack in their favour.

For allocators evaluating private credit funds with material software exposure, the practical implication of Waldron's point is clear: analysis cannot stop at lender protections and yield. It must extend to the resilience of the equity beneath. Key questions include the entry valuations at which sponsors acquired assets, the magnitude and terms of any subsequent equity injections, the extent of operational improvements realised versus pro forma, and the degree to which AI and other technological shifts may disrupt core customer value propositions. Lenders who merely look at historical default and recovery statistics for software lending without interrogating these equity dynamics risk underestimating tail risk.

Equally, for software company executives and boards, the message is that equity market discipline is already back. The era when abundant private capital would fund perpetual growth without close scrutiny has faded. As AI changes cost structures and competitive moats, and as higher rates persist, equity will continue to bear the brunt of experimentation and missteps. Private credit remains available, but is likely to price risk more discriminately and insist on clearer visibility into how equity sponsors will support businesses through volatility.

The broader significance of this perspective lies in its reminder that capital structures are ecosystems. Focusing on a single layer in isolation, whether equity or private credit, can be misleading. In software-heavy portfolios, where cash flows are promising but uncertain, and technological trajectories are in flux, the interaction between layers determines where losses land and where new value is created. By insisting that scrutiny turn first to the equity that stands in the first-loss position, Waldron is not exonerating private credit from risk; he is demanding a more sophisticated conversation about where fragility actually resides and how it will propagate through the stack over time.

References

Sonali Basak, "Goldman's AI Expectations" (LinkedIn analysis of Goldman Sachs commentary on AI and infrastructure spending).

Institutional Investor, "Goldman Expands Private Credit Ambitions With Major Overhaul".

ETF Database, "Scaling RIA Growth: The Goldman Sachs AI Playbook".

Goldman Sachs Press Release, "Goldman Sachs Announces Creation of Capital Solutions Group".

Goldman Sachs Asset Management, "Private Equity" product overview.

Goldman Sachs Asset Management, "Private Credit" product overview.

Goldman Sachs Research, "Cracks in Private Credit" (redacted report on sectoral concentration and leverage in private credit).

Private Debt Investor, "Goldman Sachs chases demand for private credit".

Term: Tertiary sales (also known as sales-out or offtakes) - FMCG / CPG

"In the FMCG/CPG industry, tertiary sales (also known as sales-out or offtakes) represent the final step in the distribution chain where an end consumer purchases a product from a retailer. Unlike primary sales (manufacturer to distributor) or secondary sales (distributor to retailer), tertiary sales reflect true market demand." - Tertiary sales (also known as sales-out or offtakes) - FMCG / CPG

Misjudging consumer demand in fast-moving categories rarely fails quietly. A misread signal at the end of the chain can cascade upstream into over-stretched factories, bloated distributor stocks, trade schemes that burn cash, and retailers losing patience with the brand. In high-velocity fast-moving consumer goods, the real constraint is not production capacity but the ability to align inventory and activity with how shoppers actually buy, store by store and week by week.

The distribution ladder: from invoice flow to consumption

In consumer packaged goods with high turnover, the supply chain is typically structured around three distinct rungs of sales measurement. At the top sit manufacturer invoices to channel partners, often referred to as primary sales; these transactions determine factory load, credit exposure and reported revenue in many organisations. One level down, secondary sales capture the movement from distributors or wholesalers into retail outlets, indicating how well the channel is absorbing stock. The final rung, and the one most tightly coupled with economic reality, is the moment when shoppers take product off the shelf and pay for it - the set of transactions described as tertiary sales, sales-out or offtakes.

This distinction matters because the three levels can move out of sync for extended periods. A manufacturer can report robust primary sales while distributors are quietly accumulating excess inventory, or retailers can be pushing units through at an accelerating pace while upstream players are slow to replenish. In both cases, the misalignment arises when decision-making is anchored on the wrong rung of the ladder. Tertiary data is the only layer that cannot be sustained by push tactics for long; it reflects the intersection of price, promotion, availability, competition and shopper behaviour in real time.

Why tertiary sales approximate true demand

FMCG and broader CPG categories are characterised by high frequency of purchase, low unit value and short shelf life for many products. The low involvement nature of these purchases means that shoppers switch easily in response to changes in price, visibility or availability. In such an environment, the volume leaving retail shelves is the closest observable manifestation of the underlying demand curve at the prices and conditions prevailing in that period.

Primary and secondary flows can be distorted by a range of mechanisms: quarter-end loading to meet targets, aggressive trade promotions that incentivise forward-buying, adjustments to credit limits, or even errors in demand planning. These interventions can inflate shipments without any corresponding change at the checkout. Conversely, tertiary sales cannot be inflated sustainably by upstream tactics; once retailer back rooms and household pantries are full, incremental shipping ceases to translate into additional offtake.

For this reason, tertiary measures often sit at the core of FMCG performance frameworks alongside market share, product penetration and share-of-wallet metrics that all depend on observed consumption rather than ship-in volumes. When category data is available at panel or scanner level, it is tertiary sales that allow brands to benchmark their performance relative to the total market, diagnose whether they are winning through penetration gains or increased loyalty, and understand the effectiveness of in-store execution.

From concept to practice: what counts as tertiary sales?

In practical terms, tertiary sales comprise all shopper purchases mediated through retail and other points of sale for a given SKU, channel and period. This may be captured through retailer EPOS data, independent scanner services, distributor sell-out tracking tools, or survey-based consumption panels. Each source brings different levels of granularity and reliability, but all aim to measure the same fundamental event: consumer offtake.

At a store level, tertiary data is often broken down by SKU, pack type and promotional status, and can be mapped against merchandising metrics such as on-shelf availability, number of facings, and share of shelf. By combining sales-out with these indicators, a brand can distinguish between an offer that is not selling despite strong visibility and one that is constrained by out-of-stock incidents or insufficient space. For instance, an SKU that sells strongly wherever it is listed but has low distribution and share of shelf indicates an opportunity for heavier listing and planogram negotiation rather than a price or concept problem.

In emerging markets, where formal EPOS coverage is limited, field sales applications and distributor management systems attempt to approximate tertiary sales by capturing order patterns from individual outlets and reconciling those with inventory movements. Although this introduces additional noise, the intent remains the same: to understand how fast stock is turning at the last mile, not merely how much has been pushed into trade channels.

Mathematical specification and linkage to other KPIs

Because tertiary sales sit at the heart of many FMCG KPIs, it is useful to formalise some of the core relationships. Let denote the tertiary volume of SKU in period , measured in units or cases. A simple identity relates stock and flows at store level:

where represents secondary sales (deliveries into the store). Rearranging highlights how tertiary sales can be inferred when stock counts and deliveries are known:

At brand level, volume market share over a given geography and period is commonly expressed as:

where the numerator aggregates tertiary sales for all SKUs belonging to the brand and the denominator sums tertiary sales for the entire product category in the same universe. A similar relationship can be written for value share by replacing volumes with sales value at retail prices.

Penetration - the proportion of households purchasing a product in a period - is another pivotal measure derived from tertiary transactions. If denotes the number of households that bought at least one unit of the brand during period , and represents the total number of households in the market, then:

From a forecasting perspective, planners often approximate future tertiary sales using time-series models where is expressed as a function of its own history, promotional flags, seasonality indicators and explanatory variables such as price and distribution. While the exact specification can vary, the goal is to predict with enough accuracy to set production and shipment plans, avoiding both stock-outs and excessive inventory.

Parameter meanings and drivers of tertiary demand

The parameters that influence tertiary sales in FMCG tend to cluster around four broad domains: consumer demand, retail execution, competitive dynamics and macro factors. Consumer demand parameters include underlying category need states, household income, demographic structure and cultural consumption patterns. Retail execution parameters cover numeric distribution, weighted distribution, on-shelf availability, shelf space allocation, price compliance and adherence to promotional plans.

Competitive dynamics introduce additional parameters such as the breadth and depth of competitor portfolios, their pricing strategies, and their promotional pressure over time. Macroeconomic variables - inflation, employment levels, input cost shocks - can compress or expand category volumes, particularly in discretionary sub-segments of CPG. A coherent tertiary sales model must either explicitly include these parameters or ensure they are absorbed by appropriate fixed effects and seasonal adjustments.

Through the lens of these parameters, practitioners often decompose changes in tertiary sales into building blocks: distribution gains, increased velocity per store, pack mix shifts, price increases, and promotion uplift. For example, if a brand grows its tertiary volume by 10 % over a year, the decomposition might reveal that 6 percentage points came from more stores carrying the product, 3 from higher units per store per week, and 1 from enhanced promotional responsiveness. This analytical discipline keeps focus on structural drivers rather than headline volume alone.

Schools of thought: primary-led vs tertiary-led management

Within FMCG organisations there are broadly two philosophies regarding which rung of the distribution ladder should anchor decision-making. A primary-led approach centres on factory shipments and distributor offtake; targets, incentives and planning are most heavily tied to these figures. This school emphasises asset utilisation, credit recovery and internal revenue recognition. It tends to dominate in environments where visibility to retail data is limited or fragmented.

A tertiary-led approach, by contrast, treats sales-out as the primary object of management. Here, the central question is not how much has been shipped, but how much is leaving shelves relative to potential. Success is defined in terms of market share, household penetration, on-shelf availability and perfect order rates, with primary and secondary flows treated as consequences of getting demand generation and execution right. Where EPOS data is rich, this philosophy can permeate everything from annual planning to daily store-level actions.

Hybrid models also exist, particularly in markets where some modern trade retailers provide detailed sell-out feeds while traditional trade remains opaque. In such cases, companies might manage modern trade by tertiary metrics and traditional trade by a blend of secondary sales and retail audit estimates, while slowly increasing the coverage of sell-out measurement tools. The tension between the two schools manifests in debates about incentive structures: should sales teams be rewarded for shipments, for offtake, or for a weighted combination of both?

Key debates and tensions around tertiary focus

One recurring debate concerns the reliability and ownership of tertiary data. Retailers may be reluctant to share detailed sell-out information, or may provide it in inconsistent formats with quality issues. Even when high-quality data is available, there can be disputes over the timing and inclusion of returns, voids or substitutions. Organisations that place heavy weight on tertiary measures must invest in robust data governance, alignment on definitions, and reconciliation processes between sell-in and sell-out views.

Another tension arises from the lag between investment and observable impact at tertiary level. Trade promotions, new distribution openings and marketing campaigns often aim to shift shopper behaviour, but their effects may take several cycles to manifest in stable offtake patterns. If tertiary metrics are used too rigidly for short-term performance evaluation, they can discourage experimentation and lead to under-investment in long-term brand-building activities that primarily influence future demand.

There is also a practical concern about attribution. Changes in tertiary sales are the net outcome of many moving parts - price, promotion, shelf execution, assortment, competition and macro conditions. Over-emphasising tertiary performance without a structured framework to dissect causes can create a culture of reactive firefighting. For example, a sudden dip in offtake may prompt hurried price discounting when the underlying issue is a competitor gaining an additional facing in a key retailer or a supply chain disruption causing intermittent on-shelf unavailability.

Finally, in some regulatory environments, the point at which revenue can be recognised for accounting purposes is tied more closely to primary sales than to tertiary offtake. Finance and commercial teams must therefore balance the external reporting requirements anchored in shipments with the internal management need to understand and influence end-consumer demand. This duality is a source of ongoing debate about which metrics should drive bonuses, budgets and strategic plans.

Tertiary sales as the organising KPI for execution

When used thoughtfully, tertiary sales provide a unifying thread connecting diverse operational and strategic KPIs. At the most basic level, store-level offtake is linked to merchandising measures such as on-shelf availability, out-of-stock rate, number of facings, and presence of point-of-sale materials. If tertiary volume is below expectation, these metrics are the first ports of call in diagnosing execution gaps: is the product present, visible, priced correctly and supported by appropriate materials?

At a higher level, tertiary sales underpin market share and penetration analyses that guide portfolio strategy. Brands can segment outlets by their offtake profiles, distinguishing high-potential stores with under-developed sales from low-potential stores that are already saturated. Field teams can then be routed and incentivised based on their capacity to unlock incremental tertiary volume in priority outlets, measured by uplift in sales-out rather than just the number of visits or orders logged.

From the finance and supply chain perspective, tertiary data supports more accurate demand planning, better inventory turnover and improved order fulfilment rates. Understanding how quickly stock is selling at retail, and how that varies by region and channel, enables more precise allocation of production and more responsive replenishment. This reduces both the risk of stock-outs, which directly suppress tertiary sales, and the risk of obsolete inventory that must be discounted or destroyed.

Why tertiary sales still matter in a digitising CPG world

The increasing digitisation of commerce and data flows might suggest that the distinction between primary, secondary and tertiary sales is becoming less salient, but the opposite is occurring. As online grocery platforms, quick-commerce players and direct-to-consumer channels proliferate, the number of nodes at which traditional FMCG products meet the end consumer is expanding. Each of these nodes generates its own variant of tertiary data: app-based transactions, subscription replenishment streams, click-and-collect orders, and more.

For brands, this explosion of touchpoints intensifies the need for a coherent view of tertiary sales that transcends channels. Without a consolidated understanding of how much product is actually being consumed, portfolio and pricing decisions risk being skewed by pockets of over or under-performance in specific routes to market. The same product may exhibit very different elasticity and promotional responsiveness online versus offline, and tertiary data is the primary mirror in which these differences become visible.

Moreover, consumer expectations of availability and service levels continue to rise, with shoppers increasingly intolerant of stock-outs or misleading online stock indicators. Meeting these expectations demands planning systems that treat tertiary demand as the starting point and work backwards to determine optimal primary and secondary flows. As sustainability concerns grow, there is also mounting pressure to minimise waste from over-production and expired stock - a task that can only be tackled effectively when the cadence of actual consumption is well understood.

In this sense, tertiary sales retain their relevance not merely as a reporting metric, but as the organising principle for how FMCG and broader CPG businesses design their supply chains, structure their incentives, and evaluate their success. The closer decision-making is tethered to genuine offtake, the less room there is for illusions created by loading the channel or chasing short-term shipment spikes. In categories defined by high velocity and thin margins, that discipline is often the difference between durable category leadership and an impressive but fragile volume story that unravels when the shelf, and the shopper, are finally heard.

Quote: John Waldron - Goldman Sachs

"Ultimately, I don't know what [AI] means in terms of our headcount and how many people are doing which jobs. I'm sure there'll be some new jobs that get developed, but one thing I'm pretty convinced of is that our really capable, smart people are going get more capable... Hopefully that means our client relationships become stickier, the value proposition becomes stronger." - John Waldron - Goldman Sachs

Large banks are treating artificial intelligence less as a discrete project and more as a pervasive change in how work is organised, priced and delivered to clients . In this shift, the most consequential decisions are not about which model to deploy, but about how to reconfigure the relationship between scarce human expertise and increasingly capable machine systems. John Waldron's comments sit precisely in this tension: the technology is advancing faster than any credible workforce forecast, yet the competitive edge of a firm like Goldman Sachs still rests on human judgement, relationships and trust, not just on compute.

In practice, this means senior leadership must hold two apparently conflicting ideas at the same time. On the one hand, there is genuine uncertainty about how many roles will be automated, reshaped or created as AI diffuses through trading, investment banking, risk, operations and wealth management. On the other hand, there is a very specific conviction: the people who are already top performers will gain disproportionate leverage from these tools, amplifying their impact with clients. The quote does not duck this contradiction; it treats workforce numbers as a contingent outcome while treating talent amplification and client value as the non-negotiable strategic horizon.

The backdrop is an industry-wide arms race. Capital markets and investment banking are data-dense domains where milliseconds, marginal insights and slightly better client service translate directly into fee pools and franchise value. Firms are experimenting with large language models, agentic systems that can orchestrate tasks across multiple internal platforms, and domain-specific models trained on proprietary research and transaction data. Waldron's framing reflects an awareness that simply swapping humans for machines misses the deeper opportunity: using AI to turn each high-value banker, salesperson, risk manager or wealth adviser into a far more productive and differentiated node in the client network.

One way to read the remark about not knowing what AI means for headcount is as a rejection of simplistic automation narratives. Banks have lived through several waves of technology-induced transformation: the move from floor to electronic trading, the rise of algorithmic execution, the digitisation of retail banking, and the first generation of robotic process automation in operations. Each time, job counts moved in complex ways. Some roles vanished, others morphed, and entirely new categories - from electronic trading quants to cybersecurity teams - emerged. AI promises a wider, deeper transformation, but its labour market effects are still non-linear and hard to predict with precision.

Yet the conviction that the most capable people will become more capable is not just comforting rhetoric. It is grounded in how AI tools actually work at scale inside a financial institution. Generative models and AI agents are particularly good at reducing search costs, summarising complex information, drafting first versions of documents, and running scenario analyses that would previously have taken days. For a high-performing banker or portfolio manager, this does not replace the job; it changes the mix of time spent between low-value tasks and high-value tasks. Instead of manually synthesising dozens of analyst reports, regulatory filings and market moves, they can spend more time on structuring bespoke solutions, negotiating, and reading subtle client signals.

In that sense, AI functions as a force multiplier on human capital. It compresses the information-gathering and analysis cycles while leaving the judgment-laden, relationship-heavy decisions to people. This is particularly true in contexts where the stakes are high and the outcomes are path-dependent: complex M&A transactions, large capital raisings, cross-border restructurings and bespoke hedging strategies. Even the most advanced models struggle with accountability and with understanding the political, cultural and interpersonal nuances that govern whether deals succeed. Senior leaders at firms like Goldman Sachs therefore have an incentive to push AI hardest into the parts of the stack where repeatable pattern recognition and calculation dominate, while reinforcing the human ownership of outcomes.

That is also where the idea of "stickier" client relationships enters the picture. In investment banking and institutional sales, stickiness is not just about price or product; it is about the client's sense that their adviser understands them better than anyone else and can move faster and more creatively on their behalf. AI, deployed well, can support this in two ways. First, it enhances institutional memory: systems can surface prior deal structures, historical interactions, risk incidents and bespoke constraints instantly, making each interaction feel more tailored. Second, AI agents can coordinate execution across trading, research, legal, risk and operations, reducing friction and error. When a client's experience shifts from cumbersome and bureaucratic to anticipatory and seamless, they are less inclined to shop around.

However, this potential is contingent on a careful balancing act between automation and authenticity. If clients perceive that their supposedly "trusted" relationship is mediated primarily by bots, templated responses or generic recommendations, trust erodes. Waldron's emphasis on making capable people more capable implicitly recognises this. The strategic bet is that AI will remain largely invisible at the surface of the relationship, embedded behind the scenes in analytics, workflow orchestration, compliance checks and personalisation engines, while the main interface remains a human team. The more that team is equipped with timely insights, contextual prompts and operational support, the more distinctive their service feels.

Factual context also matters. Goldman Sachs has been investing in technology infrastructure and data platforms for years, building the plumbing needed for more advanced AI deployments. Initiatives in transaction banking, electronic execution, consumer platforms and internal developer tooling have all contributed to a landscape where data is more accessible and systems more modular than in a traditional, siloed bank. Public commentary from the firm's leadership emphasises not only generative AI but also narrow models embedded in risk management, fraud detection and operational resilience . This context helps explain why the uncertainty is about people rather than about the technology itself: the leadership is betting that the technical foundations are already good enough to support ambitious AI use cases.

There is also an internal cultural dimension. Investment banks are talent-driven organisations, and senior management must signal to their workforce how to feel about AI. If the message were primarily about cost cutting and role elimination, the effect would be corrosive, driving away exactly the kind of high-initiative employees who are most needed to integrate AI into real workflows. By contrast, Waldron's framing positions AI as a capability enhancer for "really capable, smart people". This is not just flattery; it is a way of aligning incentives. The people most motivated to experiment with, and adopt, new tools are those who can see a direct line from that adoption to their own performance and client impact.

Yet sceptics will point out that large financial institutions are unlikely to ignore the cost-reduction potential of automation. Operations, middle-office functions, compliance checks, trade processing and certain forms of research are all ripe for productivity gains. Over time, this will almost certainly translate into fewer people doing some categories of work. The unresolved question is whether the new roles created - AI product managers inside banking divisions, data stewards, governance specialists, model risk experts, prompt engineers embedded in deal teams - will be enough to offset the roles displaced. Waldron does not pretend to know the net number, and that humility is itself telling: credible leaders today avoid making precise headcount forecasts about technologies whose adoption curves and regulatory constraints are still in flux.

Strategically, the remark hints at a broader shift from thinking about "jobs" to thinking about "tasks" and "capabilities". Even if a job title, such as equity research analyst or loan operations specialist, persists, the task composition of that job may change radically. Routine tasks are more exposed to automation; non-routine, interpersonal and judgment-heavy tasks are more likely to be augmented. From a management perspective, the right question is no longer "Which jobs will disappear?" but "Which tasks can we automate so that scarce human attention is reallocated to higher-value activities?" AI agents that can execute multi-step workflows across internal systems accelerate this decomposition of work into modular components.

For example, in a simplified depiction of a workflow, a human banker today might spend a significant share of time assembling data, populating pitch books, and manually checking constraints across risk, legal and tax. With AI agents orchestrating these steps in the background, the banker's time is reweighted towards designing creative structures, rehearsing negotiation angles, and cultivating the client relationship. The measurable productivity of that banker - deals originated per year, win rate on mandates, revenue per head - could rise substantially. From the firm's perspective, it becomes rational to concentrate more resources around such high-leverage individuals, further reinforcing the idea that the best people become even more central, not less.

This naturally raises concerns about inequality within the workforce. If AI disproportionately augments those already at the top of the performance distribution, while automating away more routine work, the result could be a more polarised organisation with a smaller middle. Junior staff might worry that the traditional apprenticeship model - learning by doing the grunt work - will erode if AI handles many of the foundational tasks. Leaders need a plan for how to maintain skill development pathways in a world where a model can generate 80 % of the first draft and a set of agents can assemble the data room overnight.

One plausible response is to redesign training and progression so that early-career professionals are explicitly taught how to supervise, critique and improve AI outputs. Instead of spending years perfecting the mechanics of spreadsheet work or drafting, they might move more quickly into roles that require critical thinking, risk awareness and client interaction, with AI handling the routine mechanics under their supervision. This could make the profession more, not less, intellectually demanding from the outset, but it also demands a higher standard of digital literacy and model awareness from new hires. Institutions that manage this transition well will likely find themselves with a more capable and adaptable talent base.

The external regulatory environment adds another layer of complexity. Banks operate under intense scrutiny regarding model risk, fairness, data privacy and operational resilience. Any large-scale use of AI, especially generative models, must be embedded in rigorous governance frameworks: model validation, bias testing, audit trails, human-in-the-loop controls, and clear lines of accountability when things go wrong. This governance overhead can slow down the deployment of AI into certain decision-making processes, keeping humans formally "in charge" in ways that preserve many existing roles. Over time, however, as regulators become more comfortable with well-governed AI systems, the frontier between human and machine decision-rights may shift, with corresponding effects on headcount.

There is also a competitive dynamic among clients themselves. Many of Goldman Sachs's corporate and institutional clients are deploying AI aggressively inside their own organisations. They will expect their financial counterparties and advisers to understand AI-driven business models, valuation dynamics, and operational risks. For a banker, salesperson or risk adviser, being "more capable" means not only using AI tools internally but also engaging credibly with clients on their AI strategy. This creates a feedback loop: the more AI reshapes the real economy, the more valuable are advisers who can straddle both finance and technology, and the more sense it makes to augment those advisers with powerful internal AI systems.

Critics might argue that narratives centred on augmenting smart people risk obscuring legitimate concerns about transparency and accountability. If decision-making becomes a hybrid of human judgement and opaque model outputs, it can be harder for outsiders - clients, regulators, even boards - to know who is responsible. The same tools that make client relationships stickier can also increase information asymmetries, as banks deploy proprietary AI to extract insights from data that clients cannot see. Addressing these concerns will require a mix of explainable AI practices, transparent governance and, importantly, a culture that treats AI as an advisory input rather than an oracular verdict.

Another objection is that emphasising "stickier" relationships and stronger value propositions could be read as a soft way of saying "higher margins" and "greater pricing power". Clients may worry that AI allows banks to segment them more finely, extract more surplus, or cross-sell more aggressively. There is some truth to this: better analytics and personalisation do sharpen commercial strategies. But the counterpoint is that sophisticated clients are not passive; they can benchmark, negotiate and multi-home across providers. In this environment, the institutions that use AI to genuinely improve outcomes - better execution quality, more resilient portfolios, more innovative deal structures - will likely be rewarded with loyalty, while those that use it primarily to squeeze clients may face backlash.

What makes Waldron's comments distinctive is the explicit refusal to present AI as a neat, linear story about job counts. Instead, the focus is on a direction of travel: towards a firm where the human core remains central but is surrounded by increasingly capable digital infrastructure, and where client relationships are deepened through better insight and execution rather than merely defended by incumbency. The underlying bet is that in a market where many players will have access to similar foundational models, the differentiator will not be the model itself but the combination of data, culture, governance and human talent wrapped around it.

For Goldman Sachs, and for its peers, the stakes are high. If they over-automate, they risk hollowing out the very human capabilities that justify premium fees and enduring relationships. If they under-invest, they risk being outcompeted by more agile players, including technology firms encroaching on financial intermediation. Steering between these extremes requires exactly the kind of nuanced view reflected in Waldron's statement: technological confidence paired with organisational humility, and a willingness to let the precise headcount outcomes emerge from a series of experiments rather than from a predetermined spreadsheet.

Ultimately, what matters is not whether the aggregate number of jobs goes up or down by some percentage, but whether the institution manages to re-architect work so that human potential is amplified rather than diminished. If AI makes the best people better, gives more clients access to that enhanced capability, and does so under robust governance, then the technology will have served as a catalyst for a new kind of financial advisory practice. If, instead, it leads to shallow automation, brittle systems and eroded trust, the promise of stickier relationships and stronger value propositions will remain unfulfilled. Waldron's remarks implicitly challenge his organisation - and, by extension, the industry - to aim for the former path, even while admitting that the journey's precise employment contours cannot yet be mapped.

References

Sonali Basak, "Goldman's AI Expectations" (LinkedIn post).

Term: Sales-in (also called Primary Sales) - FMCG / CPG

"Sales-in (also called Primary Sales) in the FMCG/CPG sector is the sale of goods directly from the manufacturer to its distributors or wholesalers. This represents the first transaction in the distribution chain, where the manufacturer officially records revenue and moves inventory out of the factory." - Sales-in (also called Primary Sales) - FMCG / CPG

Movement at the manufacturer-to-distributor stage is where demand becomes measurable in a distribution-led consumer business. The invoice raised to a distributor tells management that inventory has left the factory gate, revenue has been booked, and the next phase of the selling system must absorb that stock into the market. In FMCG and CPG, that first transaction is not just an accounting event; it is the earliest operational signal of whether a brand's channel architecture is working.

That distinction matters because fast-moving consumer goods are bought frequently, consumed quickly, and replenished often, which creates a supply chain built around high turnover rather than long ownership cycles. FMCG is often treated as a subset of CPG with a faster sales velocity, while CPG covers a broader set of consumer products with slower-moving categories as well. In that setting, primary sales, sometimes called sales-in, measure the flow of goods from the company to direct trade partners such as distributors or stockists, and they are typically tracked as the value or volume billed over a period.

What the measure captures

Primary sales describe the first commercial transfer in the distribution chain. The manufacturer invoices the distributor, transfers ownership or shipment rights according to the agreed terms, and removes the stock from finished goods inventory. What the company sees at this point is not shopper demand directly but the purchase decision of an intermediary. That is why primary sales are best understood as a channel-loading metric: they show how much product the business has pushed into its trade network, not how much has ultimately been bought off the shelf by consumers.

This makes the metric especially useful in businesses with wide geographic spread and multi-tier distribution. FMCG companies rely on extensive distribution networks because their products have high turnover, short shelf life, and repeated replenishment needs. In such systems, a strong primary sales number may indicate distribution expansion, improved distributor confidence, seasonal stocking, or an active promotion cycle. A weak number may point to cautious ordering, poor route coverage, stock overhang, payment friction, or a mismatch between planned dispatches and actual retail demand.

The practical meaning is therefore dual. On the one hand, primary sales represent official revenue recognition at the point of billing to the trade partner. On the other, they function as a proxy for the health of the supply chain, because distributors rarely keep ordering at scale unless they expect downstream movement. That is why many FMCG organisations analyse primary sales alongside secondary sales, inventory days, outlet coverage, and order frequency. Primary sales alone tell you what entered the channel; secondary sales tell you what left it.

Why it is not the same as consumer demand

The main analytical risk is confusing channel loading with true market consumption. A month of strong primary sales can coexist with weak retail take-off if distributors are building stock instead of replenishing it. The reverse can also happen: retailers may be selling through quickly while primary sales lag because distributors are already carrying elevated inventory from earlier dispatches. This is why primary sales are often described as a leading indicator rather than a complete measure of demand.

In FMCG, that distinction can materially affect management decisions. If management reads a rise in primary sales as proof of improved demand and raises production accordingly, it may unintentionally amplify inventory accumulation in the channel. If it reads a temporary dip as a loss of market share, it may overreact by cutting supply just as downstream demand is recovering. The right interpretation depends on whether the channel is under-stocked, balanced, or over-stocked, and that requires triangulating primary sales with retail off-take and distributor inventory data.

How the metric is specified

The simplest specification is the sum of invoice value generated to distributors over a chosen period: where is the number of distributor invoices in the period and is the value on the th invoice. If the analysis is done in physical units rather than value, the equivalent expression is: where denotes the quantity billed on each invoice.

When organisations compare performance over time, they often separate the value and volume dimensions. Value can rise because of price increases, pack-mix changes, or premiumisation even when unit movement is flat. Volume can rise while value stays muted if the company is discounting or shifting mix towards lower-priced packs. For that reason, a robust primary sales review normally pairs revenue-based and unit-based views, and then breaks them down by territory, SKU, channel, and distributor.

Some teams also derive growth rates using a standard change formula: where is the current period and is the comparison period. This is useful, but only if the comparison is like-for-like. In FMCG, seasonality, festival cycles, weather effects, and promotional calendars can be large enough to distort a simple period-on-period reading.

What the parameters mean in practice

In the formula, is more than a number on an ERP report. It embeds trade terms, discounts, taxes, pack structure, and the product assortment shipped to the distributor. reflects the physical loading of the network and is often more diagnostic for supply chain planning because it is less sensitive to price changes. The choice of metric depends on the decision being made. Finance teams may prioritise billed value because it maps closely to revenue. Supply chain and sales operations teams may focus on units because it relates more directly to inventory movement and replenishment capacity.

Time also matters. Primary sales can be measured daily, weekly, monthly, or quarterly, but monthly analysis is common because FMCG ordering patterns are lumpy and distributor billing can be influenced by payment cycles and month-end targets. A short interval can be noisy; a longer interval can hide emerging problems. Effective use of the metric therefore requires a cadence that matches the commercial rhythm of the business. For a fast-moving category, a monthly number may still be too coarse unless it is supplemented by weekly trend lines and route-level data.

Major schools of thought

One school treats primary sales as a financially grounded control metric. In this view, the main purpose of the number is to monitor revenue recognition, working capital release, and the speed at which stock leaves the factory. Management attention is directed towards despatch efficiency, debtor management, and channel inventory discipline. The virtue of this approach is clarity: the number is objective, billed, and easy to reconcile against accounting records.

A second school treats primary sales as a distribution health indicator. Here the emphasis is less on revenue mechanics and more on whether the channel is willing and able to absorb product. A rise in primary sales suggests distributor confidence, stronger route reach, and better alignment between company dispatches and downstream demand. This is why primary sales is frequently discussed alongside distributor productivity, outlet coverage, and range selling in FMCG performance frameworks.

A third school argues that primary sales should never be interpreted in isolation because it can be strategically misleading. This view gives greater weight to secondary sales, retail inventory, sell-through, and service levels. Supporters of this approach note that the business ultimately wins or loses at the consumer-facing shelf, not at the point of invoice to the trade. They therefore regard primary sales as necessary but incomplete, especially in markets where distributors may load stock ahead of a price rise or promotional period.

Common tensions and debates

One persistent debate concerns whether strong primary sales should be celebrated as growth or treated with caution as channel stuffing. The answer depends on whether downstream stocks are healthy. If distributors are over-ordering to exploit incentives or hedging against shortages, high primary sales can mask weak market consumption. If, however, the channel is under-stocked and service levels are poor, a lift in primary sales may simply reflect recovery to a more normal stocking position. The number only becomes meaningful when read together with inventory turnover and retail offtake.

Another tension lies between forecasting and execution. Primary sales can be boosted by aggressive sales targets, but if the underlying demand is not there, the result is bloated inventory and strained distributor relationships. Conversely, under-forecasting can leave the channel starved of stock, leading to lost sales and poor visibility on shelf. The better-managed FMCG businesses increasingly use data from distributor systems, route calls, and demand sensing tools to narrow this gap. That does not eliminate error, but it reduces the risk of mistaking inventory movement for sustainable demand.

A further debate concerns the role of incentives. Since distributors respond to margins, schemes, and credit terms, primary sales can be influenced by commercial design as much as by market pull. A temporary incentive may create a spike in invoices without a corresponding rise in consumer sell-through. This is not inherently bad, because channel build can be strategically useful before a launch or seasonal peak. The issue is whether the business knows why the spike occurred and whether the stock will convert at the retail end.

Why the measure still matters

Despite its limitations, primary sales remains a central KPI because it sits at the junction of revenue, supply, and channel behaviour. It is usually the first number that tells a manufacturer whether product is moving out of the plant and into the market system. In a sector defined by speed, repetition, and narrow operating margins, that early signal has real value. It helps sales teams see whether distributors are placing orders, helps supply chain teams plan production and dispatches, and helps finance teams judge revenue cadence.

Its importance also reflects the structure of FMCG and CPG itself. These are categories with frequent purchasing, low unit prices, and a need for reliable availability. In that environment, growth is rarely achieved through a single dramatic sale. It is built through repeated small transactions, careful channel management, and the cumulative effect of many distributor orders. Primary sales captures the first part of that chain, which is why it remains one of the most watched indicators in the sector.

As consumer markets become more data-driven, the metric is also evolving from a simple billing total into a diagnostic tool. Brands increasingly segment primary sales by SKU, outlet cluster, territory, and distributor quality, using the pattern rather than the raw total to identify emerging opportunities and risks. That makes the measure more useful than ever, but also more dependent on context. A primary sales number without channel inventory, secondary sales, and assortment data can mislead. With those inputs, it becomes a practical lens on how a consumer goods business is actually moving product through its network.

Quote: John Waldron - Goldman Sachs

"We're seeing 20, 30, 40% productivity gains in coding, in some cases, much more than that. I think the more we can develop that [AI] capability, the better we'll be... we'll be able to push more output, you know, with a relatively similar number of coders." - John Waldron - Goldman Sachs

Productivity claims on the scale now associated with generative AI would, if even half accurate, amount to one of the largest step-changes in white-collar work since the spreadsheet entered finance in the 1980s . For a bank like Goldman Sachs, whose economics are deeply leveraged to the output of knowledge workers rather than machines or factories, the idea that software development teams could reliably produce 20 to 40 percent more code, features, or shipped products without a corresponding increase in headcount is strategically transformative. It reshapes the cost base, alters the calculus of technology investment, and changes what counts as a competitive moat in financial services.

To understand the weight of this shift, it is important to see coding productivity not as a niche concern of the IT department, but as the engine room behind trading platforms, risk systems, client portals, and regulatory infrastructure. Every trading desk, risk committee, and operations team inside a modern investment bank ultimately sits atop layers of software. When senior leadership starts publicly anchoring on double-digit productivity gains from AI systems , they are signalling more than enthusiasm for a new tool; they are outlining a potential re-rating of how fast the firm can build, adapt, and respond to markets.

From incremental tools to multiplicative leverage

Historically, software productivity improvements in large banks have been incremental. Better integrated development environments, automated testing pipelines, and cloud infrastructure each chipped away at bottlenecks. But none promised to fundamentally alter the ratio between engineering headcount and shipped functionality. Generative AI coding tools, and increasingly AI agents that can manage multi-step tasks, are qualitatively different because they do not just automate infrastructure; they intervene directly in the intellectual work of design and implementation.

When leaders talk about 20, 30 or 40 percent productivity gains, they are implicitly referring to a bundle of effects: boilerplate generation, test creation, documentation drafting, code review assistance, and sometimes design suggestions. Each individually appears modest. Together, they turn into a form of multiplicative leverage: every engineer can attempt more ideas, refactor more aggressively, and support more systems without burning out. In a bank running thousands of applications, even a small reduction in friction compounds across portfolios and years.

In that sense, AI assistance functions less like a discrete tool and more like an ambient capability woven into the entire software lifecycle. Instead of one-off automation projects, you get a steady ratcheting-up of throughput across planning, coding, testing, and maintenance. That is what makes senior executives willing to talk publicly about a future in which output rises while headcount remains roughly flat: the promise of ambient leverage rather than isolated productivity hacks.

Goldman Sachs as a technology-driven bank

Goldman Sachs has, for years, framed itself as a technology company as much as an investment bank. The development of platforms like Marquee for institutional clients and Marcus in consumer banking depended on large-scale internal software capability. That positioning matters because it reveals why leadership is particularly sensitive to changes in the economics of code. When an institution already sees engineering as a revenue driver rather than a pure cost centre, a potential 40 percent improvement in software output is not just an efficiency story; it is a growth narrative.

John Waldron, as president and chief operating officer, sits at the intersection of business lines and operational capabilities. His comments about AI-enabled productivity gains are therefore not trivia from a technical specialist but guidance from someone responsible for balancing risk, cost, and strategic opportunity. When someone in that role speaks about being able to deliver more output with a relatively stable cohort of coders, he is hinting at a coming shift in the scale and ambition of internal projects as much as a possible evolution in staffing philosophy .

In a bank that spends billions annually on technology, modest percentage changes accumulate into hundreds of millions of dollars in effective capacity. The firm can either bank those gains as lower marginal cost or redeploy them into new products, better risk analytics, and differentiated client services. Historically, Goldman has often preferred to reinvest in capability rather than simply shrink, which suggests that AI productivity improvements are more likely to manifest as faster feature velocity, more experimentation, and broader platform coverage than as crude headcount reduction.

The mechanics of AI-enabled coding productivity

When executives cite specific percentages, they are usually drawing from early pilots, internal metrics, and vendor studies. In practice, productivity in software development is notoriously difficult to measure. Lines of code are a poor metric; story points and sprint velocity are context-dependent. Yet some mechanisms of improvement are sufficiently concrete to be credible.

First, AI tools reduce the time spent on repetitive or boilerplate tasks. Generating data access layers, configuration scaffolding, logging, and simple integration code can often be done in moments. Second, they accelerate developers through unfamiliar libraries or languages, by synthesising examples and suggested usage patterns quickly. Third, they assist in debugging and refactoring, offering alternative implementations or pointing to likely sources of error. Finally, they can draft tests and documentation, activities that are vital but often under-resourced in deadline-driven environments.

These are not speculative benefits. Engineers report that, when properly integrated, AI coding assistants cut the friction associated with starting work, reduce context-switching time, and make it easier to maintain focus on higher-level design. The gains are uneven across tasks and teams, but they skew heavily toward routine coding and glue work that, in aggregate, absorbs a substantial fraction of engineering hours in a complex enterprise environment.

Why stable headcount is part of the message

The claim that institutions will be able to push more output with a similar number of coders carries an implicit reassurance to employees and regulators: this is a story about augmentation, not immediate mass displacement. For a highly regulated bank, signalling that AI is a force for capability rather than indiscriminate job cuts helps maintain trust internally and externally. It also reflects a practical reality: complex financial systems cannot simply be rebuilt overnight by machines, nor can firms dispense with human oversight in areas touching client money, market integrity, or regulatory reporting.

In the near to medium term, AI-native development still depends on human engineers to frame problems, validate outputs, manage integration risks, and interpret ambiguous requirements. What changes is the ratio between time spent wrestling with syntax and time spent reasoning about system behaviour and business impact. In this environment, holding headcount roughly constant while raising expectations on throughput is a rational management posture: it keeps institutional knowledge in place while stretching teams toward higher-value work.

There is also a capital markets angle. Public statements from a top executive that output can grow without proportional increases in staff serve as a signal to investors about operating leverage. If technology spending can be kept under control while the firm expands its platform capabilities, then margins can improve without sacrificing growth. That narrative plays well with analysts accustomed to viewing banks as cost-heavy, regulation-burdened institutions struggling to differentiate themselves.

Strategic tension: efficiency vs innovation

Yet the same capability that allows a bank to deliver more with the same team can, if mismanaged, produce a culture of relentless efficiency at the expense of thoughtful innovation. When every developer can suddenly be assigned more work, there is a risk that AI productivity gains simply feed into higher utilisation targets and tighter deadlines, rather than curiosity-driven exploration. The strategic question for leadership becomes whether to frame AI as a way to strip out cost or as a means of investing in better products and resilience.

For Goldman Sachs, which competes both with large universal banks and nimble fintech firms, the temptation to use AI purely to reduce technology budgets will be balanced by the need to maintain a reputation for sophisticated, high-touch services. Pushing more output with the same headcount could mean more risk models, better scenario analysis, richer client analytics, or more real-time insights for traders. Alternatively, it could deteriorate into feature bloat and technical debt if the organisational incentives reward shipping volume over quality.

This tension is amplified by the fact that AI-generated code is not free from defects or biases. If teams move faster but do not invest proportionally in testing, observability, and governance, the bank could accumulate invisible vulnerabilities inside trading systems, pricing engines, or compliance tools. The costs of such vulnerabilities in regulated markets can be severe, including fines, reputational damage, and forced remediation programmes.

Risk, control, and regulatory scrutiny

Large financial institutions operate under intense regulatory oversight. Supervisors are already paying attention to how critical models are built, validated, and monitored. The introduction of AI into the software factory raises new questions: how is training data governed, who is accountable for AI-suggested changes that later prove faulty, and what happens when model-driven tools shape systems that themselves influence markets?

Regulators will not accept a defence that blames AI tools for coding mistakes. For all practical purposes, responsibility still rests with the bank and its human staff. As a result, the push toward higher productivity must be accompanied by equally disciplined controls: robust code review processes, traceability of changes, clear documentation of where and how AI tools are used, and internal policies on acceptable use. If AI systems are allowed to generate substantial portions of risk or pricing engines, the validation burden will rise accordingly.

There is also an emerging concern about systemic risk. If many large institutions adopt similar AI tools, trained on overlapping corpora of code and best practices, there is a possibility of correlated failure modes. Subtle bugs or design patterns recommended by these tools might propagate across institutions, creating common vulnerabilities. From a systemic perspective, productivity gains at the micro level could translate into new forms of macro fragility if they reduce diversity in implementation approaches.

Debates and objections

Not everyone accepts the headline productivity statistics at face value. Critics argue that current measurement approaches often capture short-term speed-ups in micro-tasks but underweight the costs of mis-specification, integration, long-term maintenance, and security hardening. A developer who codes faster using AI might inadvertently accept suggestions that introduce subtle performance problems or security gaps, which only surface months later.

There is also concern that productivity gains are unevenly distributed. Senior engineers who already understand system architecture may gain moderately from autocomplete and boilerplate generation; junior developers might gain more but also face a risk of becoming dependent on tools instead of learning fundamentals. Over time, this could erode the depth of expertise in the organisation, making it harder to tackle novel problems that fall outside the distribution of patterns seen during AI training.

Another objection centres on cultural dynamics. When managers hear enthusiastic numbers about 30 or 40 percent gains, some may treat them as a new baseline, rather than as an upside scenario. That can erode trust if frontline engineers feel that executive optimism is not grounded in the realities of complex legacy systems, regulatory constraints, or the sheer difficulty of coordinating large programmes of work in a bank with many stakeholders.

Why AI coding productivity matters for finance

Despite these debates, the broader significance of AI-driven productivity in software is hard to overstate for financial services. The structure of modern markets is increasingly defined by code: algorithmic execution, electronic market-making, collateral management, risk aggregation, and regulatory reporting all depend on intricate systems. The speed at which a bank can adapt those systems to new market conditions, products, or rules directly affects its competitiveness.

When senior leaders commit to scaling AI capability across development, they are effectively betting that the institution can absorb change faster than rivals. Faster deployment of risk controls after a market event, quicker rollout of client-facing tools that exploit new data sources, and more rapid iteration on trading algorithms all translate into economic advantage. Productivity in coding thus becomes a lever on the bank's agility under uncertainty, not just its cost structure.

Moreover, clients increasingly expect digital experiences on par with the best consumer technology platforms. Meeting those expectations demands continuous enhancement of interfaces, analytics, and integrations with client systems. AI-augmented development offers a way to keep pace with that escalating bar without constantly increasing headcount in an already competitive hiring market for software engineers and data scientists.

Medium-term implications for talent and organisation

One of the less discussed consequences of AI-enabled productivity is its impact on the skill profile of engineering teams. As low-level coding becomes easier, the premium shifts further toward system design, domain understanding, and the ability to translate fuzzy business needs into robust technical specifications. In a bank, that means developers who understand derivatives, collateral flows, risk methodologies, or regulatory regimes become even more valuable.

Over time, roles may polarise. Some engineers will specialise in orchestrating AI assistants across complex build pipelines; others will deepen as domain specialists who ensure that what the machines generate makes sense in the context of trading strategies, risk policies, or legal constraints. For leadership, the challenge is to redesign career paths, training programmes, and incentive structures to reflect this new division of labour, rather than treating AI tools as a simple plug-in to existing workflows.

Organisationally, the ability to produce more with the same number of coders may encourage a shift toward smaller, cross-functional teams owning end-to-end services. If each team can deliver more features per unit time, the overhead of coordinating large monolithic programmes may become less attractive. Instead, modular architectures with well-defined interfaces, owned by empowered teams, could flourish. AI assistance then becomes a force multiplier in an already agile organisational design.

Looking ahead: from pilots to structural change

For now, much of the reported 20 to 40 percent productivity improvement exists in the realm of pilots, early adopters, and selected teams. The hard work lies in translating scattered successes into structural change: standardising tools, integrating them into secure enterprise environments, training thousands of developers, and establishing governance frameworks that satisfy internal risk functions and external regulators.

That journey will not be linear. Some teams will experience impressive leaps; others will find that legacy constraints, regulatory requirements, or local culture blunt the impact of AI tools. The aggregate productivity number will likely fluctuate as the organisation learns where AI adds real value and where it introduces more risk than benefit. Metrics will need to evolve beyond simplistic measures of speed to include stability, incident rates, and client satisfaction.

Yet the direction of travel is clear. When a senior executive at a major bank speaks publicly about the ability to produce significantly more software output with stable headcount, it signals an institutional commitment to weaving AI into the core of technology operations . The specific percentages may be debated, but the strategic bet is that software, already central to modern finance, will be increasingly co-written by machines, and that the firms that learn to harness that collaboration safely and effectively will shape the competitive landscape.

In that future, productivity gains are not just about doing the same things faster. They are about enabling new kinds of systems, richer analytics, and more responsive products that would have been prohibitively expensive to build with traditional tooling. For Goldman Sachs and its peers, the real test will be whether AI-accelerated coding translates into better-managed risks, deeper client relationships, and resilient infrastructure, rather than simply a busier release calendar. The stakes extend beyond internal efficiency to the functioning of the financial system itself.

References

"Goldman's AI Expectations", LinkedIn post by Sonali Basak.

Term: Forward share - FMCG / CPG

"In the Fast-Moving Consumer Goods (FMCG) and Consumer Packaged Goods (CPG) industries, forward share refers to the percentage of total physical retail shelf space (planogram) a specific product occupies compared to its competitors." - Forward share - FMCG / CPG

Competitive advantage in grocery and convenience retail is often decided before a shopper reaches for a product. The amount of shelf space a brand secures shapes visibility, trial, substitution and ultimately sales, which is why forward share is treated as a practical measure of market presence rather than a purely descriptive merchandising statistic. In FMCG and CPG settings, the metric links physical distribution to commercial power: brands that win more shelf frontage usually gain more opportunities to be seen, picked up and bought, while rivals squeezed into thinner facings must work harder to achieve the same outcome.

Forward share is most useful because it turns a messy retail reality into a comparable percentage. It asks a simple question: what proportion of the relevant shelf, bay or fixture does one product occupy relative to the total competitive set? That makes it a merchandising KPI, but also a proxy for retailer confidence, supplier leverage and execution quality in-store. It sits alongside related concepts such as shelf share, facings and share of shelf, but its value lies in combining operational detail with strategic meaning.

What the measure captures

At its most practical, forward share describes the percentage of total physical retail shelf space a product occupies versus competing products in the same planogram. The focus is on the visible, shopper-facing area, not backroom inventory or distribution depth. That distinction matters because a product can be well stocked yet poorly presented, or prominently displayed yet at risk of running out. Forward share therefore reflects *where* the product appears in the shopper journey, not just *whether* it is available.

The metric is especially relevant in categories where products are substitutable, purchases are routine and shelf browsing is a major influence on choice. FMCG and CPG businesses often sell low- to mid-priced items bought frequently and with limited deliberation, so physical placement can have a disproportionate effect on sales conversion. Broader industry commentary on FMCG and CPG emphasises rapid turnover, repeated replenishment and retail distribution as defining characteristics, which is precisely why shelf allocation remains commercially important.

Why retailers and brands care

Forward share is important because shelf space is finite and expensive in strategic terms. Retailers use planograms to organise assortment, manage category flow and maximise sales per linear metre, so every extra facing allocated to one brand is a decision about the rest of the category. For suppliers, forward share is a negotiable outcome of brand strength, trade terms, promotional support, historical performance and retailer confidence. For analysts, it is a shorthand for how much physical presence a brand has won compared with its competitors.

The operational relevance is straightforward. A stronger forward share can improve visibility, reduce the chance of shoppers overlooking a product and support conversion when consumers are making a quick choice. It can also reinforce brand legitimacy: a product that appears repeatedly across a fixture can seem more established or popular than one with a marginal display. In practice, this means shelf presence can influence demand as much as it reflects it, creating a feedback loop between commercial success and retail allocation.

How the percentage works

The calculation is conceptually similar to market share, but the denominator is shelf space rather than sales. If is the forward share of product , is the shelf space or facings assigned to that product, and is the total shelf space assigned to the competitive set, then the basic expression is . Here is the number of products, and the result is expressed as a percentage of the visible shelf block under review.

This formulation is simple, but the measurement choices are not. Shelf space can be counted by linear centimetres, by facings, by shelf depth or by a weighted view that accounts for positioning at eye level versus lower shelves. Some retailers and field teams use face count because it is quick to audit, while others prefer linear space because it better captures true physical presence. The key point is consistency: forward share only becomes meaningful when the same method is applied across stores, categories and time periods.

Measurement choices and parameter meanings

In analytical terms, the variable definitions matter more than the arithmetic. If is measured as facings, then the KPI captures how many product fronts are visible to shoppers. If is measured as shelf length, then the KPI better reflects actual retail real estate. If the denominator excludes out-of-stock facings or temporary promotional ends, the number tells a different story again. Because the metric is comparative, any inconsistency in the measurement boundary changes the interpretation.

The most important practical parameters are the category scope, the competitive set and the store format. A cereal brand's forward share in a large supermarket is not directly comparable with its share in a convenience store, because the total fixture, assortment depth and shopper mission differ. Likewise, a retailer-owned display may inflate one brand's physical presence temporarily without altering its baseline planogram position. For this reason, forward share should be read as a context-specific indicator rather than a universal law of brand strength.

Relationship to market share

Forward share is often discussed as a physical analogue to market share, but the two are not identical. Market share measures the proportion of category sales captured by a company or brand, usually by revenue or volume, whereas forward share measures the proportion of the shelf captured in the store. The resemblance is useful because both express relative dominance, yet the causal direction can run both ways. Higher sales may justify more shelf space, but more shelf space may also drive higher sales by improving visibility and choice probability.

This is why retailers and suppliers use the metric in both retrospective and prospective ways. Retrospectively, it helps explain why a brand gained or lost sales. Prospectively, it supports decisions about range rationalisation, assortment expansion and planogram redesign. In that sense, forward share is not merely descriptive. It is an allocation signal embedded in the merchandising process, and it can be used to test whether shelf investment is aligned with commercial performance.

Major schools of thought

One school of thought treats forward share as a pure execution KPI. On this view, the main question is whether the shelf allocation agreed in the planogram has been implemented accurately in store. The focus is compliance: do the actual shelves match the intended design, and is the brand receiving the space it was promised? This approach is common in field sales, retail audits and outlet standards monitoring, where the number is used to identify planogram breaches or distribution errors.

A second school of thought treats forward share as a strategic KPI. Here the measure is less about operational compliance and more about competitive positioning. A brand with a larger shelf presence than its sales justify may be seen as over-invested, while a brand with a smaller shelf presence than its sales support may be seen as under-distributed. This perspective is often used in category management, where the objective is to optimise the whole shelf for category growth rather than maximise any one supplier's interests.

A third school of thought links forward share to shopper behaviour. In this view, shelf space is a behavioural intervention: it changes the probability of notice, comparison and purchase. The metric therefore matters because it may influence choice architecture, not simply because it records a commercial bargain. This interpretation is especially relevant in high-frequency, low-involvement categories where shoppers rely on visual heuristics and time-saving cues.

Tensions and debates

The central debate is whether forward share measures cause, consequence or compromise. If a brand already sells well, does it deserve more shelf space because demand is proven, or does the additional shelf space create the demand? In practice, the answer is often both. Retailers may allocate shelf based on historical velocity, but once the allocation changes, future sales can shift in response. That makes the KPI useful, but also vulnerable to circular reasoning if analysts forget the allocation itself can shape the outcome.

Another tension concerns the retailer's and supplier's interests. Suppliers may argue for more forward share to improve visibility and support growth, while retailers must protect category productivity, shopper convenience and margin. A supplier-led view can overstate the importance of a single brand, whereas a retailer-led view may understate the role of brand equity and shopper pull. The most credible category decisions usually sit between those positions: shelf space should reflect sales performance, margin contribution, strategic role and shopper mission, not just negotiated power.

Quote: John Waldron - Goldman Sachs

"We're not replacing all of our people with digital agents, but we're going expect you to understand how to work with a digital agent? Or if you're in the front office part of the firm talking to a client, we're going expect you to understand how to use that tooling to make your relationships stronger and to deliver a better value proposition." - John Waldron - Goldman Sachs

The immediate pressure inside large financial institutions is not a cinematic wave of robot bankers, but a subtler requirement: every human professional is being nudged to become a hybrid operator who can orchestrate digital agents alongside traditional judgement . In investment banking, markets, and wealth management, the constraint is no longer computing power; it is how quickly relationship managers, product specialists, and risk teams can retool their daily workflows so that artificial intelligence becomes a multiplier rather than a parallel system they quietly ignore.

This shift reflects a hard commercial edge. If a rival bank can prepare pitches, price scenarios, and risk views in a fraction of the time because its staff are fluent in agentic tooling, the competitive gap compounds quickly. In primary issuance, the team that can synthesise cross-asset signals, simulate client outcomes, and draft tailored communication overnight will win mandates that slower teams never see. Waldron's message, framed as an expectation rather than an option, signals that the cultural battle is no longer about whether AI will be used, but about who learns to drive it well enough to matter in front of clients .

The meaning behind "not replacing all of our people"

On its surface, the reassurance that people are not being wholesale replaced sounds like standard corporate calming language. In context, it is closer to a restructuring of what counts as baseline competence. For a long time, elite finance rewarded those who could manually process information faster than peers: building models at speed, scanning research, remembering client constraints. As digital agents become competent at these tasks, the advantage migrates toward those who can specify problems, structure prompts, check outputs, and integrate them into client-facing decisions.

In practice, that means the benchmark analyst or associate no longer differentiates themselves by staying later to scrub spreadsheets. The organisation is explicitly saying that survival requires learning how to shape, interrogate, and supervise systems that perform the mechanical work. The phrase "we're not replacing all of our people" is therefore less a promise of job security and more a statement that replacement will be selective and correlated with an individual's ability to collaborate with the new tools rather than compete against them.

For senior leaders, this reframes talent strategy. Instead of assessing whether someone can master an existing process, they now need to ask whether this person can translate messy, client-specific questions into forms that a digital agent can process reliably, and then explain the augmented answer back to a client in plain language. Those who cannot bridge that gap risk becoming the human bottleneck in an otherwise modernised workflow.

Why front-office AI fluency matters more than back-office automation

Back-office automation has long been an easier sell: it cuts costs, reduces errors, and rarely touches the firm's image with clients. Front-office AI is far more sensitive. When a president and chief operating officer signals that client-facing staff must understand how to work with digital agents, he is betting that competitive advantage now lies in visibly augmented advice and responsiveness, not only in silent efficiency gains behind the scenes .

Consider how a traditional relationship manager prepares for a meeting with a corporate treasurer. Historically, they might pull recent trading history, read internal research, and ask a product team for structured ideas. With agentic tools, the same banker could have a digital assistant that ingests the client's historical exposure, peer behaviour, current market conditions, and regulatory constraints, then generates scenario-specific talking points, risk flags, and tailored documentation drafts. The human still leads the relationship, but the quality and breadth of information they bring to the table are transformed.

The strategic tension is that such augmentation subtly changes what clients expect. Once a few banks routinely show up with richer insights and faster follow-ups thanks to AI assistance, the old level of preparation begins to look negligent. Front-office AI fluency therefore becomes not just an internal efficiency play, but an external signalling device: a proof that the institution is at the frontier of information use on the client's behalf.

From "using a tool" to working with an agent

Language matters here. A tool is something you operate directly; an agent is something you delegate to. The move from simple interfaces to digital agents implies workflows where staff specify goals, context, and constraints, then monitor an autonomous or semi-autonomous system as it executes tasks across multiple data sources and applications.

Technically, these agents may chain calls to large language models, internal APIs, pricing engines, and document stores, deciding step by step which resource to invoke next. Conceptually, the employee's role shifts toward being a supervisor and editor. Rather than manually compiling a client memo, they might instruct an agent: assemble current exposure, relevant research, regulatory developments, and three hedging strategies with quantified trade-offs. The agent generates a draft; the banker then checks, corrects, and contextualises it before sharing.

This shift raises questions of accountability. If an agent suggests a structure that later underperforms or creates unforeseen risks, how will the firm trace the reasoning? In markets, where internal models already embody layers of opacity, adding another AI-driven layer intensifies the need for auditability. Institutions like Goldman Sachs must therefore build governance frameworks in which every agent interaction is logged, reproducible, and attributable, so that human supervisors remain genuinely accountable rather than nominally in charge of a black box.

The quantitative core: productivity, risk, and value

Although Waldron's statement is framed in everyday language, the underlying trade-offs can be expressed in simple quantitative terms. Suppose a banker's effective output without AI is represented by , and the introduction of a well-used digital agent yields a proportional productivity gain , where . The new output is . If a firm targets, for example, , each banker must deliver more insight, coverage, or revenue at similar or lower risk just to meet expectations.

From a risk perspective, let baseline decision quality (probability of making a correct or acceptable recommendation) be . Introducing an AI agent with independent error probability does not automatically improve outcomes. If the banker blindly follows the agent, the new error probability may be worse than . In contrast, if the banker uses the agent as a second opinion and only acts when human judgement and agent output agree, a simplified model of agreement-based acceptance could be written as: error probability , where represents the chance of misinterpreting or mis-implementing a correct suggestion. The firm's training and process design effectively aim to minimise while maximising output .

Over a portfolio of relationships, expected value can be thought of as , where is client revenue, is servicing cost, and is expected loss from errors or misconduct for relationship . Digital agents are introduced to increase by enabling more tailored solutions, reduce by automating labour, and contain or reduce through better risk detection. The danger, and the core of the internal debate, is that poorly governed AI may reduce but increase by enabling faster, more systematic mistakes.

Strategic context inside Goldman Sachs

Goldman Sachs is not approaching AI from a standing start. The firm has long invested in quantitative research, systematic strategies, and technology-heavy businesses. What is changing is the expectation that AI literacy will spread beyond quant desks and engineering teams into the heart of client coverage. Waldron's role as president and chief operating officer gives his words operational force: they hint at performance assessments, promotion criteria, and training programmes being redesigned around AI fluency as a core competency .

In a firm whose brand rests on high-touch advisory relationships, this introduces a delicate balancing act. On one hand, digital agents promise to scale the reach of top talent: ideas from a leading sector banker or risk specialist can be embedded into prompts and templates used across hundreds of colleagues. On the other hand, there is a reputational risk if clients perceive interactions as scripted, generic, or driven more by machine output than by genuine understanding of their unique situation.

To manage this, Goldman must distinguish between invisible augmentation and visible automation. Invisible augmentation quietly improves the quality, timeliness, and depth of what human bankers bring to meetings. Visible automation, by contrast, risks making conversations feel standardised or mechanised. Waldron's emphasis on using tooling to make relationships stronger, rather than simply more efficient, implicitly recognises this tension and anchors the strategy in client experience, not just internal cost metrics.

Internal debates and cultural resistance

Within any large bank, reactions to such expectations fall along a spectrum. Younger staff may be eager adopters, viewing AI agents as a way to compress years of manual apprenticeship into a faster learning curve. Mid-career professionals may worry that their hard-won pattern recognition is being devalued or that they will be judged on unfamiliar technical skills. Senior rainmakers may insist that relationships are fundamentally about trust, implying that technology is peripheral.

One internal debate concerns whether mandating AI use risks creating over-reliance. If every pitch deck, market commentary, or client note flows through a digital agent, the danger is that staff gradually lose the habit of constructing narratives and checking numbers themselves. That could hollow out the talent pipeline, leaving the firm with a generation of bankers who are adept at editing but less capable of building original analysis from first principles.

Another debate centres on fairness and performance measurement. If some teams gain early access to powerful agents while others are limited to basic tools, comparisons of productivity or revenue generation become distorted. The firm must therefore decide whether to standardise agent capabilities across divisions or tolerate a period of uneven adoption while experimenting. In a competitive, bonus-driven culture, these choices carry real political weight.

Objections from clients, regulators, and staff

Clients may voice concerns that their confidential data is being fed into opaque systems, potentially used to train models or inform strategies for competitors. To reassure them, institutions need clear boundaries around data use, on-premise or private-cloud deployments, and the ability to explain where and how AI is applied in servicing a particular account. Without that, assurances about stronger relationships may ring hollow.

Regulators, for their part, are already alert to the risks of model-driven decision-making. When client-facing staff rely on digital agents for suitability assessments, product recommendations, or pricing, questions arise about explainability, bias, and control. Supervisory authorities may require firms to demonstrate that AI-assisted processes comply with existing conduct, suitability, and market abuse rules, even if the rules were written long before such systems existed.

Staff themselves may object on ethical grounds, worrying that AI-enabled surveillance of their work will intensify. Because digital agents often sit within monitored platforms, every prompt, draft, and correction can be logged and analysed. This creates opportunities for training and quality control, but also a sense that the margin for informal experimentation is shrinking. Waldron's expectation that employees learn to use such tools sits against this backdrop of expanded visibility.

Capability building: from training to institutional memory

For a mandate like this to take hold, the firm must invest heavily in capability building. Generic AI literacy courses will not suffice. Relationship managers need domain-specific playbooks: how to brief an agent about a mid-market industrial client versus a sovereign wealth fund; what prompts yield useful early-stage M&A idea screens; how to constrain an agent to only use pre-approved language when drafting sensitive communications.

Over time, the knowledge embedded in these playbooks may become an asset in its own right. As staff experiment, refine prompts, and correct outputs, the firm can capture these interactions, distil best practices, and encode them into institutional templates. In effect, the organisation starts to build a second layer of institutional memory residing not only in human experience and documentation, but in the configuration of its digital agents.

This, however, raises governance questions similar to those that arose around spreadsheets and internal models decades earlier. Who owns a particularly effective prompt structure: the individual banker, their team, or the firm? How are changes to shared agent templates tested and approved? If a flawed prompt leads to a systematic error in client materials, at what point does responsibility shift from the individual user to the central team that maintains the agent framework?

Why this expectation matters beyond Goldman Sachs

When a leading global bank tells its people that working with digital agents is no longer optional, it sends a signal across the industry. Competitors must either match the expectation, risking cultural backlash of their own, or accept that their staff may be less augmented in critical client interactions. Smaller institutions and buy-side firms will watch closely, weighing whether they can emulate the approach without the same scale of technology investment.

Labour markets will also respond. Candidates entering finance increasingly face an implicit test: not just whether they understand accounting, valuation, and markets, but whether they can demonstrate practical fluency with AI-assisted workflows. Business schools and professional training programmes are likely to adapt their curricula accordingly, blending traditional financial theory with hands-on experience in specifying, supervising, and critiquing digital agents.

In the longer term, the expectation that every client-facing professional can work with a digital agent accelerates a broader transformation in how expertise is distributed. Some forms of know-how that were once scarce and localised may become widely accessible through well-designed agents. What remains scarce is the ability to orchestrate that know-how in complex, high-stakes human situations: to choose when to lean on the agent, when to override it, and how to translate its suggestions into actions that preserve trust.

A quiet redefinition of professional competence

Underneath the reassuring language about not replacing people lies a redefinition of what it means to be a high-performing professional in finance. Competence now includes an ability to collaborate with systems that are probabilistic, opaque in places, and evolving. Relationship strength is no longer measured only by how well a banker remembers a client's history or instincts, but by how effectively they can harness institutional data and AI tooling to anticipate needs and deliver tailored solutions.

In that sense, Waldron's statement is less about technology and more about identity. The banker, trader, or advisor who sees themselves solely as an individual expert risks obsolescence; the one who sees themselves as the conductor of a small orchestra of digital agents, data sources, and human colleagues is closer to the emerging norm. As these expectations spread, the boundary between human judgement and machine assistance becomes less a line of defence and more a design problem: how to structure work so that each does what it is uniquely good at, in service of relationships that are, paradoxically, meant to feel more human than ever.

References

Sonali Basak, "Goldman's AI Expectations" (LinkedIn analysis of John Waldron's remarks).