Identifying Profit Pools and Strategic Bottlenecks in Value Chains

Hello. In the previous lesson, you mapped customers, complementors, suppliers, channels, and competitors around a specific offering. That map showed who depends on whom. This lesson adds two questions: Where does operating profit accumulate across those participants? And which participants control something others cannot readily replace?

We will build a profit-pool map for a fictional AI-enabled enterprise workflow, then use software and semiconductor examples to distinguish a lucrative position from a strategic bottleneck. Allow about 40–45 minutes, including the selected resources.

Set the boundary before looking for profit

Your earlier Product, Growth, Distribution, People, and Enabling Functions map described activities inside a company. An industry profit-pool map crosses company boundaries. For an AI-enabled workflow, it might examine compute equipment, cloud capacity, model services, applications, and implementation services. Those are not five departments of one firm; they are potential places where different firms earn revenue and incur costs.

Keep the two maps separate, then connect them. For instance, access to enterprise customers may be a scarce resource in the industry map. Inside an application company, the work of securing that access may sit largely in Growth and Distribution. A costly internal activity does not automatically constitute an industry profit pool, and an industry's most profitable position need not be the activity with the most employees.

Choose a scope narrow enough for meaningful comparison: a customer workload, the competing ways to serve it, a set of participating businesses, and a time period. “AI” is too broad. “Enterprise AI-assisted customer-service workflows over a defined year” is more workable. For a platform, include important complementors and transaction partners; a single firm's internal chain will otherwise miss where value is created or captured.

Umbrex’s Porter’s Value Chain explains how to scope and quantify an activity map, and why platform analysis needs an accompanying view of partners and control points.

A useful boundary test is to ask whether each slice represents a distinct economic activity whose revenue and profit you can estimate. “Chip design” and “cloud hosting” can qualify. “A large technology company” usually cannot: it may participate in several slices at once.

Measure the pool, not the prominence of the player

A profit pool is the operating profit earned across the activities in a defined industry scope. Its distribution matters as much as its total. A stage can have immense sales but thin margins; a smaller stage can earn more profit by retaining a larger share of its revenue.

Think Insights’ Profit Pools introduces the concept and a practical mapping sequence. Read its discussion of why profit varies within an industry before its five-step method.

The basic calculation for each slice is:

“Normalized” matters. A one-off gain, an unusual restructuring charge, or a temporarily subsidized price can obscure the economics you are trying to compare. Use the same period and a consistent profit definition across slices. EBIT is often a workable starting point, though a capital-intensive activity may look less attractive once its invested capital is considered.

The supplied How Profit Pool Mapping Works image is a compact checklist. Notice its emphasis on choosing the metric, aligning slice definitions, and showing absolute profit, not just margin.

Consider the following fictional annual estimates for companies participating in an enterprise AI-workflow market. They are illustrative assumptions, not estimates for actual firms.

The application slice contributes $24/58.4$, or about 41% of measured EBIT, despite having neither the highest margin nor a monopoly on customer relationships. The model-services slice earns only $3 million in this illustration: technical sophistication alone does not establish that it captures the most profit.

There is an accounting trap here. A cloud provider may sell to a model provider, which sells to an application provider, which sells to an enterprise. Do not sum the five revenue figures and call the result end-customer spending: some payments appear again as costs at downstream firms. The table sums participants’ operating profits within a stated scope, not distinct dollars spent by final customers. Check whether any firm appears in more than one slice, allocate its activity consistently, and eliminate any double-counted profit. For marketplaces, also distinguish gross transaction value from the platform’s own revenue, such as fees.

For a first-pass exhibit, show revenue as each slice’s width and operating margin as its height; the resulting area represents absolute operating profit. A negative-profit slice must remain visible rather than disappearing from the chart. Reconcile your slice estimates against another view—such as company-segment profits—then label unresolved differences and assumptions. Precision should reflect the quality of the underlying data.

Find the bottleneck—and ask who can monetize it

A strategic bottleneck is an activity, asset, interface, or capability on which an important outcome depends and for which timely substitutes are limited. Scarcity alone is insufficient. The strategic question is counterfactual: If this participant changed its price, terms, capacity, or access rules, what realistic alternatives would others have, and how quickly could they use them?

Bloomberg Originals’ How AI Is Pushing the Semiconductor Supply Chain to the Limit gives two concrete places to apply that test. Treat its examples as evidence of supply dependence at the time described, not as a substitute for measuring current profit.

A bottleneck diagnosis becomes stronger when it specifies five things:

1. Dependency: Which downstream products or participants require the input or access?
2. Substitutes: Can buyers switch technologies, suppliers, routes to customers, or product designs?
3. Time: Would switching take weeks, years, or a new generation of investment?
4. Control: Who sets capacity, compatibility rules, prices, or access terms?
5. Appropriation: Can that controller retain economic returns after its own costs, investment needs, bargaining with buyers, and competitive responses?

The last question prevents a common analytical leap. A fabrication plant can be indispensable yet capital-intensive; its owner may need sustained investment to preserve its position. Conversely, a participant with modest physical assets may control distribution or a standard and capture attractive profit. Bottleneck, bargaining power, accounting margin, and return on invested capital are related—but not interchangeable.

In the fictional AI-workflow pool, compute equipment could be a bottleneck if unavailable capacity delays every other slice. Yet the table alone cannot prove that: it gives profit estimates, not substitution times or capacity evidence. The application slice could instead control customer access through embedded workflows and integrations—but only if customers face meaningful disruption when switching. Both claims require investigation.

Test how the pool could move

Profit pools are snapshots; strategic value lies partly in anticipating their movement. A platform adds another complication: it can make an exchange more valuable while deliberately charging one participant group little, or spending heavily to attract that group. High transaction volume therefore does not establish high platform profit.

Professor Michael Cusumano discusses monetization, subsidies, and the danger of assuming that scale necessarily fixes weak economics in The Business of Platforms. His examples are historical; use the reasoning rather than treating every company assessment as a current verdict.

Return to the fictional AI-workflow map. Suppose compute supply expands and model services become easier to substitute. Equipment prices or cloud margins might compress. Profit might migrate toward applications that retain customer demand, or toward integration firms if deployments remain difficult. Equally, application margins could compress if several vendors offer comparable workflows. These are alternative hypotheses, not predictions derived from the original table.

The same method applies across the three company types you are studying:

• In enterprise software, investigate whether profit rests in a differentiated application, an installed workflow, or control of procurement and distribution. Test each against switching behavior and credible substitutes.
• In a multi-sided platform, identify who controls access or rules, but measure fees net of incentives, trust costs, and other operating expenses before declaring a profit pool.
• In an AI-native business, distinguish temporary scarcity in an upstream input from durable control of customer demand. Ask who benefits—and who loses—if that input becomes cheaper.

A concise strategy exhibit can therefore have two layers. The first sizes revenue, margin, and absolute profit by slice, with ranges where evidence is weak. The second marks dependencies and control points, naming their controller, plausible substitutes, time to bypass, and evidence for pricing power. Where the layers do not coincide—an essential but low-return supplier, or a profitable participant without durable control—you have found a question worth pursuing.

Key takeaways

To locate a profit pool, define a specific industry boundary, estimate comparable operating profit by economic slice, and reconcile the result without mistaking upstream sales or platform transaction volume for final-customer revenue. To locate a bottleneck, test dependency, alternatives, bypass time, and control—then ask whether its owner can actually retain attractive returns. Neither the largest revenue slice nor the most essential input is automatically the best economic position.

Next, you will reconstruct a company’s segment-level business model from filings, earnings materials, and operating metrics. That evidence will help replace the illustrative ranges used here with defensible estimates of where particular firms participate and how they earn.