Opportunity recognition, applied to segment evaluation, means testing each candidate customer group against three conditions before you spend money on it: the segment must have a problem the product actually solves, the members must be able to act on that problem, and you must be able to reach them at a cost the margin can absorb. The method below turns that into a scored screen. It works because it forces you to reject segments on evidence rather than enthusiasm β and enthusiasm is the resource founders have in surplus.
Defining Opportunity Recognition as a Segment Filter Rather Than a Brainstorming Exercise
Opportunity recognition is the entrepreneurial skill of spotting a mismatch between what a group needs and what existing options give them β before the market prices that mismatch in. In practice, the academic treatment most founders encounter comes from Scott Shane's work on the individual-opportunity nexus, which argues that opportunities are not lying around waiting to be noticed; they are discovered by people with prior knowledge, and the same information produces different conclusions in different heads.
That has a hard implication for segment work. Your own industry experience determines which segments you can even see clearly. If you spent six years in restaurant supply chains, you will spot a purchasing-pain opportunity in independent restaurant groups that a former app developer sitting next to you will walk straight past. Shane's point isn't that you're smarter; it's that your knowledge base is a lens, and lenses have blind spots as well as focus.
The distinction that matters for evaluation: a segment you recognize as underserved is a hypothesis, not a conclusion. Recognition generates the candidate list. Validation kills or keeps each candidate. Founders who skip the second step tend to describe opportunity recognition as intuition, which is another way of saying they never wrote anything down.
So treat the term operationally. For this article, opportunity recognition means the structured comparison of a candidate segment's problem severity, ability to buy, reachability, and competitive slack against the same measures for every other candidate. Four dimensions, scored, comparable. That definition is narrow enough to act on.
The Three Conditions Every Segment Must Pass Before It Counts
Before scoring, run a pass/fail gate. A segment survives only if all three conditions hold; if any one fails, don't score it, discard it. This prevents a segment with a screaming problem and no budget from scoring deceptively well on a weighted average.
Condition one β the problem is painful and frequent. Not "would be nice." The test is whether a member of the segment is already spending money, time, or workarounds on the problem today. If nobody is paying for a bad solution, nobody will pay for a good one. A restaurant owner hand-keying supplier invoices into a spreadsheet every Sunday night is evidence; a restaurant owner saying "yeah, that's annoying" is not.
Condition two β the segment can transact. Can the individual you're talking to actually sign a check, or does every purchase require a committee, a procurement cycle, and a security review? Small businesses and solo professionals usually transact alone. Mid-market firms often don't. Enterprise almost never does on first contact.
Condition three β you can reach them at a viable cost. This is where most optimistic segment lists collapse. Reachability means a named channel that reliably delivers your message β a trade association, a specific publication, a search term with real volume, a distributor β and a cost per acquired customer that leaves margin on the table.
A segment that fails condition three is not a bad segment; it's a segment you can't afford yet. Write that down separately. Founders routinely discard a genuinely good segment because they lacked the channel, then rediscover it a year later and rebuild the entire plan around it.
Building the Candidate List Without Contaminating It
The candidate list is where opportunity recognition either gets disciplined or turns into a wish list. Keep it to five to eight segments, each described in customer terms rather than product terms. "Independent coffee shops with two to five locations" is a segment. "The hospitality vertical" is a category, and categories can't be scored.
Generate candidates from four sources, and label which source each one came from. First, your own prior work β the buyers you've watched struggle. Second, inbound signals: who has already emailed, called, or filled out a form. Third, adjacent migration: who currently uses a competitor's product badly enough to complain in public. Fourth, the structural gap: which group's problem is getting worse for a reason that isn't going away, such as a labor shortage or a regulatory change.
Label the source because it predicts reliability. Inbound-signal segments carry revealed preference, which is worth more than any survey. Prior-work segments carry your own bias, which is worth less than you think. Keep a column in the spreadsheet recording the source so the scoring discussion can weight accordingly.
The contamination risk is real. Once a founder has a favorite segment, every scoring weight quietly gets tuned to make that segment win. The defense is to fix the weights and the scoring rubric before you research any segment, in writing, with the date on it. If you change a weight later, note why. A rubric that changes after the data arrives isn't a rubric.
Scoring Problem Severity on a One-to-Five Scale With a Concrete Test
Severity is the dimension founders overrate most, because they mistake their own fascination for the customer's urgency. Anchor the scale to observable behavior rather than opinion.
| Score | Anchor: what you can observe | Example signal |
|---|---|---|
| 5 | Already paying for a workaround, and it's failing | Paid for a tool, abandoned it, still doing the task manually |
| 4 | Already paying for a workaround that mostly works | Competitor product in active use, complaints logged publicly |
| 3 | Spending significant unpaid time on the problem | Two or more hours per week of staff time, described unprompted |
| 2 | Complains about it when asked directly | Agreement in an interview, no spending, no workaround |
| 1 | Acknowledges it only after explanation | "I guess that's true" |
Score on the anchor, not the vibe. A segment scoring 5 has members who can describe their workaround and its cost without prompting; that's a much stronger signal than enthusiastic agreement with your framing. If you cannot place a segment on the table because you have no observation, the honest score is "unknown," and unknown segments should be researched before they're ranked.
Cap your claims here. This article gives you the rubric, not market data β you supply the observations. Two hours of customer conversations per segment, with three to five members each, is enough to place most segments on this scale with reasonable confidence. Fewer than three conversations means you're guessing.
The failure case worth naming: a segment can legitimately score 5 on severity and still be worthless because the problem is severe but rare. Frequency belongs in the severity score too. A problem that hits once a year, no matter how painful, produces a purchase cycle too slow to build a business on unless the contract value is very large.
Testing Ability to Buy: Budget, Authority, and Timing in One Pass
Ability to buy decomposes into three questions you can answer in a single conversation with a segment member. Does a budget line exist for this kind of purchase? Who signs? How long from first conversation to money moving?
Run it as a script. Ask what they currently spend on the problem area, who approves a purchase of your likely price point, and what the last similar purchase timeline looked like. The answers give you budget, authority, and timing without requiring a formal survey.
| Segment type | Typical budget holder | Typical approval path | Signal to watch |
|---|---|---|---|
| Solo professional | Themselves | None | Buys on a card the same week |
| Small business, under 20 staff | Owner | Owner, sometimes with a bookkeeper's input | Decision within days to a few weeks |
| Mid-market, 20β500 staff | Department head | Head plus finance, often a vendor onboarding process | Six to twelve weeks |
| Enterprise, 500+ | Procurement | Security review, legal, committee | Three months to a year |
The table is a pattern, not a law. Confirm the actual path in the conversation β the schema.org-style assumption that every mid-market firm runs a twelve-week cycle will cost you deals with the ones that don't. The failure case: a founder prices for mid-market, builds a sales process for enterprise, and starves because the segment's real approval path was the department head's own card.
Score the dimension on the shortest realistic path to first revenue. A segment where one person can say yes and pay this week scores 5. A segment requiring a security review before a pilot scores 2, regardless of the size of the eventual contract. If your runway is shorter than the sales cycle, the contract size is irrelevant.
Calculating Reachability as a Cost per Conversation, Not a Channel List
Reachability is the dimension that converts a segment from interesting to fundable, and it's the one most segment lists never quantify. The calculation is simple: pick the single best channel for each segment, estimate the cost to generate one qualified conversation through it, then compare that to the margin on a typical deal.
Cost per qualified conversation = total channel spend Γ· number of qualified conversations generated. Qualified means the person matched your segment definition and engaged in a real conversation about the problem. Not a click. Not a download.
Worked example with figures you can verify yourself: if a targeted trade-publication sponsorship costs $4,000 and historically produces 20 qualified conversations, the cost per conversation is $200. If a typical deal in that segment is worth $3,000 in first-year margin and one in five conversations closes, the acquisition cost per customer is $1,000 against $3,000 of margin β workable. If the close rate is one in twenty, the acquisition cost is $4,000 and the deal is underwater. Same segment, same channel, different verdict once you do the arithmetic.
You won't have historical conversion data on day one. Substitute your best estimate and mark it as an estimate, then replace it with real numbers after the first cohort. The point of the calculation isn't precision; it's that it forces the channel assumption into the open where it can be argued with.
The edge case that breaks this: segments reachable only through a gatekeeper, such as a trade body that controls access to its member list. The cost per conversation looks fine until the gatekeeper's fees, exclusivity terms, or slow approval process are included. Score gatekept segments one point lower than the arithmetic suggests, and find out what the gatekeeper actually charges before you commit.
Mapping Competitive Slack: Where Incumbents Have Left the Problem Half-Solved
Competitive slack is the gap between what the segment needs and what existing options deliver. It's the dimension where opportunity recognition earns its name, because slack is precisely the mismatch the term describes. Score it by examining what segment members actually use today, including spreadsheets, WhatsApp groups, and interns.
Build the comparison row your competitors' pages never build: for each segment, list the incumbent option, the specific thing it does badly, and whether that failure is structural or temporary. Structural failures β the incumbent's business model prevents them from serving this segment profitably β create durable slack. Temporary failures β a bug, a slow quarter, a price increase β create a window you can't rely on.
| Slack type | What it looks like | Durability |
|---|---|---|
| Structural | Incumbent's pricing model can't serve small accounts profitably | High; incumbent won't chase it |
| Segment blind spot | Incumbent's product assumes a workflow this segment doesn't have | Medium-high; costly for incumbent to fix |
| Service gap | No onboarding, support, or implementation for this segment | Medium; copyable |
| Temporary | Recent price rise, outage, or public misstep | Low; closes within months |
Score 5 when the incumbent's model structurally excludes the segment. Score 3 when the product simply doesn't fit well. Score 1 when the segment is well served and merely under-informed. A segment can be well served and still be a good target if you can reach it cheaper β but that's a reachability argument, not a slack argument, and it belongs in a different column.
One caution on this dimension: "no competitors" usually means "no market," not "blue ocean." If a segment has real severity and real budget and nobody is selling to it, the most likely explanation is that someone tried and the unit economics failed. Establish why the gap exists before you treat it as opportunity.
Running the Weighted Screen: A Worked Scoring Example Across Three Segments
Fix your weights before scoring. A reasonable starting set for a first product: problem severity 30 percent, ability to buy 25 percent, reachability 25 percent, competitive slack 20 percent. Adjust to your situation β a business with a long runway can weight severity higher; a bootstrapped one should weight reachability higher, because speed to revenue is survival.
The arithmetic: weighted score = (severity Γ 0.30) + (buying Γ 0.25) + (reachability Γ 0.25) + (slack Γ 0.20), with each dimension on the one-to-five scale. Maximum is 5.0.
| Segment | Severity | Buying | Reachability | Slack | Weighted score |
|---|---|---|---|---|---|
| Independent coffee shops, 2β5 sites (estimate) | 4 | 5 | 3 | 4 | 3.95 |
| Regional bakery chains, 10β30 sites (estimate) | 5 | 3 | 4 | 3 | 3.85 |
| Catering-only operators (estimate) | 3 | 5 | 2 | 5 | 3.65 |
The figures above are illustrative estimates, not research findings β replace every cell with your own observed scores before acting on the ranking. What the example shows is the mechanics: the highest-severity segment does not automatically win, because buying friction and slack pull it down. The catering segment has the most slack and the easiest buyer but the worst reachability, which is exactly the kind of trade-off a weighted screen surfaces and a gut-feel ranking hides.
Read the ranking as a shortlist, not a verdict. The top two segments are close enough that the difference is inside your estimation error. Take both into a validation round β ten conversations apiece β and let the conversations break the tie. The screen's job is to stop you from validating five segments at once, which no early-stage team has the hours for.
Choosing Between a Weighted Screen and a Sequential Knockout When Data Is Thin
Two structures work, and they fail in different conditions. The weighted screen above suits founders with some observation across all four dimensions. The sequential knockout suits founders with almost no data, because it refuses to average unknowns.
In a knockout, you test one condition at a time in order of cheapness to check. Severity first, because three conversations answer it. Then buying, because one more conversation usually settles authority and timing. Reachability requires a small spend to test. Slack requires research. Any segment that fails a gate exits immediately, and you never spend the expensive research on it.
| Method | Best when | Fails when |
|---|---|---|
| Weighted screen | You have partial data on all dimensions | A high score on one dimension masks a fatal gap on another |
| Sequential knockout | Data is thin and research budget is limited | Segments fail early on noise rather than real disqualifiers |
The knockout's weakness is noise sensitivity: a segment can fail the severity gate because you happened to interview three unusually satisfied members. Guard against it by requiring two independent observations before a fail, and by revisiting failed segments once per quarter rather than burying them permanently.
Choose one method and write down which one you used. Switching methods mid-evaluation is how founders unconsciously engineer the answer they wanted β the knockout quietly drops the segment the founder dislikes, the weighted screen quietly weights toward the favorite. The written record is the defense.
Turning the Screen Into a Weekly Validation Loop With Named Kill Criteria
The output of the screen is a ranked shortlist plus a validation plan. The plan needs three things: who you'll talk to, what would change your mind, and when you'll decide.
Define kill criteria in advance. For the top segment, an example: if fewer than four of ten conversations reveal an active workaround, severity was misjudged β drop the score by two and re-rank. If fewer than two of ten can name the person who approves the purchase, the buying score was wrong. Writing the criteria before the conversations means the data can actually overturn the ranking.
Run the loop weekly. Ten conversations per segment per week is achievable for a founder doing nothing else; five is realistic alongside building. Log each conversation against the four dimensions immediately, because memory reconstructs conversations to fit the hypothesis within about 48 hours. A shared spreadsheet with one row per conversation and four score columns is enough infrastructure.
Then re-rank monthly, not daily. Segment scores move slowly, and re-ranking every time a conversation goes badly produces whiplash. Monthly gives you roughly forty conversations per segment between re-ranks, which is enough for the numbers to mean something.
The kill criteria deserve one more note: they must be falsifiable by a specific number. "If the segment isn't responding" is not a kill criterion. "If cost per qualified conversation exceeds $500 after two channel tests" is. The number is what stops the founder from moving the goalposts when the results turn disappointing.
FAQ
How many customer segments should a startup evaluate before picking one?
Five to eight is the practical range for a first screen, because fewer than five rarely produces a real comparison and more than eight exceeds what a small team can research in a month. Each segment needs at least three conversations to score honestly, so eight segments means twenty-four conversations before the ranking means anything. Beyond eight, founders start scoring from assumptions.
Does a high severity score alone justify targeting a segment?
No β severity can be high and the segment still unusable if the members can't transact or you can't reach them affordably. The weighted screen deliberately caps severity at 30 percent for exactly this reason. A severity-5 segment with buying at 1 and reachability at 2 scores 2.75, which ranks below a moderately painful segment that buys on a card. Pain without a purchase path is a charity case, not a market.
What's the difference between opportunity recognition and market segmentation?
Market segmentation splits a broad market into groups by shared characteristics; opportunity recognition evaluates whether a specific group's unmet need is worth acting on. Segmentation is descriptive and can be done from a database. Recognition requires observation of behavior β what people spend, what they work around, what they complain about in public. You can segment a market perfectly and still recognize no opportunity in it.
How do I estimate cost per qualified conversation with no historical data?
Pick one channel, set a small fixed test budget, and count qualified conversations strictly β segment match plus a real problem discussion. Divide spend by conversations. Two channel tests at a few hundred dollars each give you two data points, enough to rank segments roughly. Treat the first estimate as provisional and replace it after the first cohort of customers.
What happens if two segments score within a few points of each other?
Treat the gap as inside your estimation error and validate both in parallel with ten conversations each. The tie usually breaks on buying friction β one segment's members sign faster, and speed to revenue resolves the ranking better than any additional scoring. Don't add a fifth dimension to force a winner; that's rubric-tuning dressed as rigor.
Can a segment I can't reach today become reachable later?
Yes, and this is common β a channel may not exist until a trade association forms, a publication launches, or a platform opens an API. Record unreachable segments with the specific channel gap named, and revisit quarterly. The trap is building the product for an unreachable segment on the assumption the channel will appear; build for reachable demand and keep the unreachable list as a watchlist.
Is "no competitors" in a segment a positive slack score?
Usually not. Absent competitors in a segment with real severity and real budget most often means earlier attempts failed on unit economics, not that the opportunity is unclaimed. Score slack on the incumbent's structural inability to serve the segment, not on the absence of any vendor. Investigate why the gap exists before scoring it as opportunity.
How often should the segment ranking be recalculated?
Monthly is the right cadence for most early-stage teams, with roughly forty conversations per segment between re-ranks. Weekly re-ranking overreacts to single conversations; quarterly re-ranking lets a bad ranking run for three months of wasted outreach. Recalculate sooner only if a kill criterion trips, and note in writing which criterion fired.
The scoring weights, channel economics, and approval-path patterns described here are general educational information for planning purposes, not a market forecast or investment recommendation β segment scores depend entirely on observations you supply. Verify approval cycles, pricing, and channel costs with the specific buyers and providers in your region before committing budget. If your decisions involve regulated financial, legal, or health products, consult a locally qualified professional.