Choose the segment for which you have the strongest combination of urgent need, credible evidence, practical access, attractive economics and an ability to deliver. Do not simply choose the segment with the largest theoretical market. Score both candidates against the same criteria, show the evidence behind every score, and test the few assumptions that could reverse the decision.

That is the short answer. The discipline lies in separating what the team knows from what it hopes, then making an explicit, reversible decision. A scorecard helps because it exposes disagreement and trade-offs. It does not turn judgement into fact.

Start with a decision, not a market-sizing exercise

The question is not, "Which segment could one day be worth more?" It is, "Which segment should receive our next concentrated period of product, sales and marketing effort?" Add a time horizon and a consequence. For example: choose one segment for the next two quarters, fund one acquisition motion, and agree what evidence would justify reconsidering the choice.

That framing matters because market size is only one input. The UK government's guidance on understanding market size and segments recommends examining customers, demand, competition and the part of the market a business could realistically reach. The US Small Business Administration similarly places demand, market saturation, location and pricing inside market research rather than treating population size as a sufficient answer.

Write the decision in one sentence before scoring. If the sentence does not identify the two alternatives, the period of commitment and the resources affected, the team is probably debating an abstract preference rather than making an operating choice.

Make the two segments comparable

Both candidates must be defined at the same level of specificity. "Healthcare" versus "independent dental practices with three to ten sites" is not a fair comparison. One is a broad sector; the other combines industry, business model and scale.

Use the same dimensions for both:

  • the person or organisation with the problem;
  • the situation that triggers the need;
  • the job they are trying to complete;
  • the alternative they use now;
  • the buying conditions, including budget and authority;
  • the geography and any relevant constraints.

This definition is not merely demographic. The Strategyzer Value Proposition Canvas separates a customer profile into jobs, pains and gains, which is useful here because two apparently similar organisations can be different segments if the triggering job or buying process differs.

If you cannot describe both candidates using the same fields, stop and repair the definitions. Otherwise, the scorecard will reward whichever segment was described more generously.

Use this weighted segment scorecard

Set the weights before anyone reveals a preferred answer. Score each segment from 1 to 5, where 1 means materially weak and 5 means materially strong. Record one evidence note beside every score. Calculate weighted points as weight x score / 5, producing a total out of 100.

CriterionWeightA score of 1 meansA score of 5 meansEvidence to inspect
Problem urgency20The problem is tolerated or rarely triggeredThe problem creates a time-bound consequenceObserved workarounds, missed deadlines, escalations, switching events
Evidence strength20The case rests mainly on internal beliefSeveral independent sources support the same findingInterview notes, behavioural data, lost-deal reasons, paid tests
Reach and access15Buyers are hard or costly to identify and reachA repeatable route to qualified buyers already existsList quality, response patterns, partner access, sales-cycle entry points
Economic potential20Willingness and ability to pay are doubtfulPlausible value, budget and acquisition economics alignCurrent spend, approved budgets, sales effort, retention conditions
Delivery fit15Serving the segment requires major exceptionsExisting capabilities solve the important job wellImplementation effort, support load, integrations, regulatory demands
Strategic learning10Winning teaches little that transfersThe segment produces valuable proof or capabilitiesReference value, product learning, adjacent opportunities

The weights above are a starting point, not a universal truth. A bootstrapped service business may give more weight to access and cash generation. A venture-backed product team may give more weight to repeatability and learning. What matters is that the weights reflect the current decision and total 100.

Grade the evidence behind each score

A precise-looking number can conceal fragile evidence. Add a confidence grade to every score:

  • A - observed: repeated behaviour, transactions or product use directly support the score;
  • B - reported: several relevant people independently describe the same need or constraint;
  • C - indirect: credible secondary evidence supports the idea, but your segment has not been tested;
  • D - assumed: the score is based mainly on team belief or analogy.

Government user-research guidance recommends beginning with the full context of what users are trying to achieve, rather than starting from a preferred solution. Its guidance on taking research notes also distinguishes observations from researchers' interpretations. Apply both principles to the scorecard: capture what happened, then record what you infer from it.

Do not average away a critical weakness. A segment with a high total but a D-grade score for buying authority is not necessarily the leader. It is the leader only if that assumption survives a suitable test.

Run the scoring session without manufacturing agreement

Give each decision-maker the definitions, criteria and evidence pack in advance. Ask people to score independently before the meeting. During the session, discuss the largest scoring gaps first. A difference between 2 and 5 usually reveals a disputed assumption, a different interpretation of the segment or evidence that not everyone has seen.

For every disputed item, ask:

  1. 1 What specific claim does this score make?
  2. 2 What evidence supports it?
  3. 3 What evidence would change it?
  4. 4 Does the uncertainty affect the decision enough to test now?

Keep both the original individual scores and the agreed score. The pattern of disagreement is useful. If sales rates access highly while delivery rates fit poorly, the decision may require an operating change, not a debate over arithmetic.

Test the assumptions that could change the winner

Do not respond to every unknown with a large research programme. Identify the two or three low-confidence assumptions with the greatest potential to reverse the ranking.

The Strategyzer Test Card is useful because it asks a team to state the hypothesis, the test and the success threshold before collecting results. GOV.UK's guidance on planning user research likewise advises turning assumptions into research questions that support an upcoming decision.

Match the test to the claim:

  • To test urgency, observe recent examples and ask about the last occurrence, not general interest.
  • To test access, run a small, time-boxed outreach experiment using the channel you expect to scale.
  • To test buying conditions, map a real recent purchase, including approvers, procurement and budget source.
  • To test delivery fit, conduct a scoped implementation review using an actual environment or workflow.
  • To test willingness to commit, ask for a proportionate next action rather than a favourable opinion.

Define the threshold before running the test. Otherwise, a team can reinterpret ambiguous results to protect its preferred segment.

Illustrative example: a choice that changes after confidence is considered

Illustrative example - these figures are hypothetical and are not To-Do Growth research.

A workflow software company compares regional accountancy firms with specialist construction consultancies. The accountancy segment scores 78 out of 100; construction scores 74. At first glance, accountancy wins. However, its high economics score assumes partners will fund a firm-wide rollout, and that assumption has a D confidence grade. Construction's slightly lower score is supported by recent paid pilots and observed deadline failures.

The team does not declare construction the winner or ignore the scorecard. It runs two tests: five accountancy buying-process interviews focused on the last software purchase, and a small offer test that requires a partner-level discovery meeting. If the budget and authority assumptions hold, accountancy remains first. If they do not, the evidence-adjusted choice changes.

The lesson is not to subtract an arbitrary "confidence penalty" from every score. It is to treat a close result with one decision-critical assumption as unresolved until that assumption is tested.

Interpret the result and set a review rule

Use three possible outcomes:

  • Clear lead: one segment leads on total score and has no untested, decision-critical weakness. Commit for the defined period.
  • Conditional lead: one segment leads, but one or two assumptions could reverse the result. Run targeted tests before full commitment.
  • No meaningful lead: totals are close and evidence quality is similar. Choose the segment that offers the cheaper, faster or more reversible learning path.

Also run a sensitivity check. Change one disputed weight within a plausible range. If a five-point shift makes the winner flip, the scorecard is showing a genuinely close strategic choice, not a robust answer.

Record a review trigger rather than reopening the debate every week. Suitable triggers include a completed test, a defined number of qualified opportunities, a material regulatory change or the end of the agreed focus period. This gives the team permission to act without pretending the decision is permanent.

Common scorecard failures

The tool becomes misleading when a team:

  • scores a broad market against a narrow niche;
  • treats interview enthusiasm as proof of purchase;
  • uses the same evidence to justify several criteria without saying so;
  • changes weights after seeing which segment wins;
  • hides veto conditions, such as a missing certification or integration;
  • includes strategic preference as if it were customer evidence;
  • interprets a small numerical difference as certainty.

The OECD's competition assessment materials are a useful reminder that market conditions and competitive constraints deserve explicit examination. For a small business, that means the scorecard should include real alternatives and access conditions, not just attractive descriptions of customer need.

Limitations and trade-offs

A weighted scorecard formalises judgement; it does not discover truth. Criteria can overlap, scores are sensitive to definitions, and founders may still favour evidence that confirms an existing preference. Small research samples can reveal mechanisms and language but cannot establish population prevalence. Economic estimates are especially fragile before repeatable sales exist. Use the tool to expose assumptions, choose tests and document a time-bounded decision. Do not present the total as a statistically validated forecast.

Further reading

Sources

Your useful next step

Write the decision sentence, define both segments using the same six fields, and ask each decision-maker to score independently. Circle every high-weight score supported only by C- or D-grade evidence. Choose the smallest credible test for the two assumptions most likely to reverse the result, and set the date on which the decision will be made.

Publication note

Publication note: This article provides a decision method, not market, financial or statistical advice. The weights and hypothetical example are illustrative. Teams should use evidence from their own customers, economics, delivery constraints and jurisdiction before committing resources.