Research-Driven Opportunity Sizing

“How big is the opportunity?” looks like a request for a number. It is usually a request for confidence.

Should we invest now? Which segment deserves attention? Is the upside large enough to accept the risk? What must we learn before making a larger commitment?

A polished total addressable market slide can answer none of those questions. A rough, transparent model can answer all of them if it connects evidence to the decision.

Opportunity sizing is not prediction theatre. It is a disciplined way to show what could create value, what limits our ability to capture it, and which uncertainty matters next.

What a useful size should answer

An opportunity estimate should help someone choose between actions. Before gathering data, write down:

• the decision the estimate will inform;
• the alternatives being compared;
• the timeframe;
• the user and business outcome;
• the investment or risk under consideration.

“Size the market for team software” is too broad. “Decide whether to fund discovery for a compliance workflow used by mid-sized European teams” is workable.

The second question gives the research boundaries. It also prevents a global market statistic from being presented as evidence for a narrow product bet.

If the estimate will not change a decision, do not build a larger model. Clarify the decision first.

This is part of a wider Discovery & Validation practice: reduce the uncertainty that blocks the next responsible commitment.

Market size and product opportunity are not the same

Market size describes economic activity or potential demand across a defined market. It can provide context, but it does not tell one team what it can create or capture.

A product opportunity is narrower. It combines at least four dimensions:

1. Problem size: How many people experience the problem, and how consequential is it?
2. Solution fit: How much of that problem could the product credibly resolve?
3. Strategic advantage: Why is this team well placed to create the value?
4. Capture: Which share can the organisation reach, serve, and convert into a meaningful result?

A large problem with weak reach may be a poor near-term opportunity. A smaller problem inside an existing workflow may be attractive if the team has access, trust, capability, and favourable economics.

Total addressable market, serviceable available market, and serviceable obtainable market can help frame a commercial opportunity. They become dangerous when the smallest circle is still based on an arbitrary percentage.

“If we win 1%” is not a capture model. It is a wish expressed as multiplication.

Define the outcome before the population

Teams often begin with “How many users?” before agreeing what will improve for those users.

Define the opportunity as a change:

For [population], reduce or increase [meaningful outcome] from [baseline] within [timeframe], while respecting [constraint].

The baseline can be behavioural, financial, operational, or experiential. It might be time lost, failed completions, avoidable cost, churn risk, or an unmet job.

The outcome protects the model from counting people who technically belong to a segment but do not experience the relevant problem.

Then define the population with inclusion and exclusion rules. “Small businesses” is not a usable denominator. Geography, size, workflow, technology, buying authority, and problem frequency may all matter.

Write the boundary in words before looking for a number. Otherwise, the easiest available dataset will quietly define the opportunity for you.

Build an evidence ladder

No single research method can support the whole estimate. Use different evidence for different parts of the model.

Establish the population

Official business statistics, regulatory registers, product data, and credible industry sources can help establish how many people or organisations meet the boundary.

The US Census Bureau’s County Business Patterns and Eurostat’s European business statistics are examples of public sources for business counts and characteristics.

Check the definition, year, coverage, and unit. “Establishments,” “enterprises,” “accounts,” and “people” are not interchangeable.

Understand problem incidence and severity

Interviews reveal how the problem appears, what triggers it, and what people do today. Surveys or product data can then test how common those patterns are in a larger, relevant population.

Do not pitch the solution while trying to measure the problem. People can be enthusiastic about an idea they would never prioritise, buy, or adopt.

How to Approach Customer Discovery Systematically offers a deeper method for gathering this evidence.

Test solution fit

Concept tests, prototypes, concierge delivery, and behavioural experiments can test whether a proposed response changes the outcome.

Interest is weaker evidence than action. A positive interview reaction may justify the next test; it rarely justifies an aggressive adoption assumption.

Test reach and capture

Map how the population can be reached, who decides, which alternatives exist, and what blocks adoption. Sales conversations, channel tests, procurement research, and pricing work belong here.

For an existing product, historical activation, retention, and expansion data can provide useful bounds. For a new market, those rates are assumptions until evidence says otherwise.

Model the opportunity as a chain

A simple model is often more useful than a comprehensive one because people can see where the result comes from.

For a commercial account-based opportunity, a starting structure could be:

eligible population
× reachable share
× problem incidence
× expected value per account
× realistic capture rate
= captured opportunity over the chosen period

For a product outcome, use units that match the decision:

affected users
× addressable frequency
× expected change per occurrence
= potential outcome change

Do not mix units. Time saved, revenue, active users, and risk reduction can all matter, but converting them into one figure requires explicit rules.

Separate potential value from captured value. The product may create value for a customer without the business capturing all of it, and the business model may capture revenue before durable customer value exists.

Keep effort and cost outside the opportunity estimate, then compare them in the decision. An attractive opportunity can still be a poor investment if it requires disproportionate time, capability, or risk.

Use ranges instead of manufactured certainty

Every estimate rests on assumptions. Hiding them behind one point estimate does not remove uncertainty; it merely makes the model harder to challenge.

Use a range for variables that are genuinely uncertain. Label the basis of each range:

• observed in current product data;
• derived from an external dataset;
• supported by research;
• borrowed from a comparable case;
• judgement with little evidence.

Calculate conservative and optimistic cases. Add a central case only when evidence supports a defensible central estimate. The labels should reflect evidence, not mood.

Sensitivity analysis asks which assumption changes the answer most. HM Treasury’s Green Book guidance treats sensitivity and risk as part of appraisal because results depend on uncertain inputs.

If a small change in capture rate destroys the case, capture deserves research before a large investment. If the decision is robust across the plausible range, more precision may not be worth the delay.

How to Approach Assumption Mapping Systematically can help rank the beliefs behind the estimate by importance and evidence.

A hypothetical worked example

The following numbers are invented only to demonstrate the method. They are not market data or a forecast for a real product.

Imagine a workflow product considering a new compliance feature for a defined group of organisations.

The team models:

• Eligible population: 1,200 organisations
• Reachable share: 25–40%
• Problem incidence: 50–70%
• Annual captured value per adopted account: €960–€1,920
• First-period capture among relevant reachable accounts: 10–20%

The conservative case is:

1,200 × 25% × 50% × €960 × 10% = €14,400

The optimistic case is:

1,200 × 40% × 70% × €1,920 × 20% = €129,024

The useful output is not “the opportunity is €71,712,” the midpoint of those values. That number would imply knowledge the team does not have.

The useful conclusion is that captured value per account and capture have the widest relative ranges. Reach also matters, while the eligible population is relatively well established.

The next work is therefore not a more detailed population report. It is direct research on value and willingness to pay, followed by a small test of capture with qualified buyers.

Suppose the investment under consideration is larger than even the optimistic captured value. The model has already improved the decision. The team can stop, change scope, or seek a different source of value.

Suppose the conservative case still justifies a discovery sprint. The team can proceed without pretending the optimistic case is a promise.

Where sizing goes wrong

The denominator is impressive but irrelevant

A broad population is multiplied by a small invented share. The final number looks modest enough to be believable, but the share has no connection to reach, need, or adoption.

Replace the percentage with a chain of observable conditions. If those conditions cannot be researched, the estimate should remain explicitly speculative.

Optimism compounds

Each variable uses a plausible upper bound, and the model multiplies them together. Individually defensible assumptions produce a collectively unlikely outcome.

Build cases consistently. A conservative case should not pair low costs with high conversion merely because both help the conclusion.

Research samples only enthusiasts

The team interviews current power users or people who requested the feature. Problem incidence and willingness to adopt are then generalised to the whole population.

Include indifferent users, lost prospects, non-users, and people with competing workflows. The opportunity includes the resistance, not only the demand.

Solution interest replaces problem evidence

People say an idea is useful, so the team assigns a conversion rate. The estimate skips frequency, urgency, switching effort, budget, and alternatives.

Use interest to plan the next test, not to declare captured value.

Structural barriers disappear

Regulation, procurement, integration, data access, switching costs, and operational change can reduce reach or delay value.

Represent them in the model or the risk assessment. A barrier does not vanish because the spreadsheet lacks a row for it.

Effort is hidden inside prioritisation

Two opportunities are compared by size alone. The larger one wins even though it requires a new capability, a long migration, and a different sales model.

Size first, then compare value, effort, confidence, strategic fit, and timing. How to Prioritise provides a wider decision framework.

Turn the estimate into a decision

Present the model as an argument someone else can inspect, not as a final number to admire.

A useful opportunity brief contains:

1. Decision: What choice is being made now?
2. Outcome: Whose situation should change, and how?
3. Boundary: Which population and timeframe are included?
4. Model: How do the variables connect?
5. Evidence: What supports each important input?
6. Range: What changes across plausible cases?
7. Sensitivity: Which assumption controls the conclusion?
8. Recommendation: Proceed, research, reshape, defer, or stop?

Keep a record of the estimate and revisit it after delivery. Compare what happened with what the team assumed.

Calibration is the long-term advantage of sizing. A team that records its misses learns whether it habitually overestimates reach, underestimates adoption friction, or confuses interest with value.

The goal is not to become perfect at forecasting. It is to become less surprised for the same reasons.

Working artefact: the Sensitivity Ledger

Keep the ledger beside the model. It prevents a polished output number from hiding the few uncertain inputs that actually control the investment decision.

Input	Evidence state	Plausible range	Effect on decision	Owner	Next evidence action
Population	Sourced, observed, inferred, or assumed	Low, base, and high	Does the recommendation change across the range?	Named person	Better source or boundary check
Reach	Current channel evidence or assumption	Low, base, and high	Does acquisition remain feasible?	Named person	Channel test
Adoption	Behavioural evidence or analogy	Low, base, and high	Does value survive realistic friction?	Named person	Prototype or workflow test
Value	Observed outcome or stated willingness	Low, base, and high	Does the upside justify the commitment?	Named person	Outcome or pricing evidence
Constraint	Known barrier or unresolved dependency	Probability or impact range	Could it delay or eliminate capture?	Named person	Regulatory, technical, or operational check

Sort the rows by decision sensitivity, not by ease of research. The next action should reduce the uncertainty most capable of changing the recommendation.

An opportunity-sizing checklist

Before using the estimate, ask:

• Is the decision explicit?
• Does the outcome describe real value rather than an output?
• Is the population defined before it is counted?
• Are market size and capturable opportunity separated?
• Can every important variable be traced to evidence or a labelled assumption?
• Are ranges internally consistent?
• Which assumption is the result most sensitive to?
• Have reach, alternatives, constraints, and time been included?
• Is the example real, sourced, or clearly labelled hypothetical?
• What is the next smallest action that reduces decision-relevant uncertainty?

A good opportunity size does not end an argument with authority. It improves the argument by showing where evidence is strong, where judgement enters, and what the organisation is truly betting on.

That is enough. The purpose of the model is not to predict the future. It is to make the next decision more responsible.

Sources

• Market research and competitive analysis — US Small Business Administration
• European business statistics — Eurostat
• County Business Patterns — US Census Bureau
• The Green Book 2026 — HM Treasury

Read next

How to Approach Assumption Mapping Systematically helps decide which belief in an opportunity model deserves evidence first.