Step 2: From validated problem to sharp audience

Outputs doc: outputs.md, Step 2 section. Fill it in as you work through the steps below. Raw interview captures go in captures.md. Step 3 reads the Step 2 section before starting.

Jump to: Diagnostic · Step 2.1 · Step 2.2 · Step 2.3 · Step 2.4 · Step 2.5 · Step 2.6 · Step 2.7 · Step 2.8 · Summary · Assumption sweep

What this step is: The specifics of the audience Step 1 identified. Step 1 named who has the problem; Step 2 narrows that group to the sharpest segment and characterises it. Not demographics. Behavior, decision process, and context: which decision this person has to make, and what is currently keeping them from making it. The output is an audience you could build a list of right now, not a persona with a stock photo and a name.

Why it comes after Step 1 and before everything else: Step 1 told you a real problem exists and named who has it. Step 2 sharpens that group into a single segment and works out the specifics, precisely enough to reach them. If the audience is vague, Step 3 cannot write a value proposition that beats the alternative (because “the alternative” depends on who is choosing), Step 4 cannot find the language that lands (because language is specific to a situation), and Step 5 cannot pick channels (because channels follow from where this audience actually is). A broad audience here produces fuzzy work in every step below. If you cannot define a sharp audience, the usual cause is that Step 1 is not actually finished. Go back before you push forward.

What finishing this step produces: A written audience definition in the Step 2 section of outputs.md built around a behavior and a situation, naming the decision the person faces and the blocker that stalls it, with a chosen sharpest segment, sized bottom-up against a viability threshold, and, for B2B, a map of who else is in the buying decision. Backed by primary evidence, not inference.

Diagnostic: is Step 2 actually done?

Before building anything, answer these questions in writing. Vague answers mean the step is not finished.

1. Could you build a list of ten real people or companies who fit this audience right now? Not “marketers” or “mid-market companies.” Ten specific names you could find this week. If you cannot, the audience is still a category, not an audience.

2. Is the audience defined by what they do and the situation they are in, rather than by demographics or firmographics alone? “CMOs at companies with 200 to 500 employees” is a filter, not an audience. It describes who they are on paper, not why they have the problem. A real audience definition names a behavior or situation that makes the problem acute for them specifically.

3. Can you name the specific decision this person has to make, and what is currently stopping them from making it? The audience exists at the moment of a decision. If you cannot state the decision and the blocker, you know who has the problem but not what is keeping them stuck, and Step 3 will have nothing to push against.

4. Do you know who else is involved in that decision? In B2B, almost nobody decides alone. If you cannot name the others in the buying group (who feels the consequence, who controls budget, who can veto), you are likely marketing to one person while a different person holds the outcome.

5. Can you point to the sharpest segment within the audience and say why they feel the problem more acutely than the rest? A problem is felt unevenly. If everyone in your audience feels it equally in your description, you have not looked hard enough. The sharpest segment is where Step 3 through Step 6 should aim first.

6. Do you know, in defensible numbers, whether the sharpest segment is big enough to be worth winning? Not a research firm’s market size. A bottom-up count of accounts that fit your signals, a plausible price anchored on what the problem costs, and a realistic share of the segment you could win. If you cannot say what winning this segment would be worth per year, you do not know whether it is a beachhead or a puddle.

If you answered all six clearly and in writing, Step 2 may already be done. Jump to the Step 2 section of outputs.md, fill in the fields, check the checklist, and move to Step 3. If not, work through the steps below.

Step 2.1. Pull the seed audience from Step 1

Duration: 30 minutes

Step 2 does not start from a blank page. It starts from the seed audience Step 1 named in its final step. The fastest route to a sharp audience is to take that group and read your Step 1 evidence about them as people rather than as a problem.

What to do:

1. Open the Step 1 section of outputs.md and copy the following into the “Seed audience from Step 1” field in the Step 2 section:

   • The “Who” line from your final problem statement.
   • The named workaround. The people currently using the workaround are usually your best early audience, because they have already admitted the problem is worth spending effort on. Step 1 told you this; Step 2 acts on it.
   • The consequence range from your interview synthesis. Who felt the problem acutely and who barely felt it? That spread is the raw material for segmentation in Step 2.4.
   • Any sub-segments your Step 1 synthesis surfaced (different workarounds often signal different sub-groups).

2. Read it back and write one sentence: “Based on Step 1, the people who have this problem most acutely appear to be the ones who ___.” Do not polish it. It is a starting hypothesis, and the rest of this step tests and sharpens it.

3. If the “Who” from Step 1 is broad or you are pulling it forward from inference rather than evidence, that is not a Step 2 problem to paper over. Note it in the Step 2 scope notes in outputs.md and log it in assumptions.md (Step: 2, Status: Untested). A broad Step 1 “Who” caps how sharp Step 2 can get. If it is too broad to work with, the honest move is to return to Step 1.

Step 2.2. Describe the audience by behavior, not demographics

Duration: 45-60 minutes

Demographics and firmographics describe an audience but do not explain it. Age, title, company size, and industry are filters you can apply to a list, and they matter for reaching people later, but on their own they do not tell you why someone has the problem or whether they feel it. Two CMOs at two identical companies can be in completely different situations. The thing that separates the person who has your problem acutely from the person who does not is almost always something they are doing or something happening in their situation, not something on their business card.

What to do:

1. In the Step 2 section of outputs.md, under “Audience by behavior,” describe the audience in terms of:

   • What they are doing. The activity, role, or project that brings the problem into their day. “Running outbound sales without a follow-up system” is a behavior. “Being a sales manager” is a title.
   • The situation or trigger. What is happening in their world that makes the problem live right now? A new hire, a missed target, a tool that stopped working, a season, a funding round. Pull this from your Step 1 situation triggers.
   • Observable signals. The things you could see from the outside that tell you a specific person is in this audience. They have posted about the problem, they use a particular workaround tool, they have hired for a role that exists because of the problem, they are at a stage where the problem is structurally inevitable. This list is what makes “build a list of ten right now” possible, and it feeds directly into Step 5 targeting.

2. Write the observable signals as a checklist a stranger could apply. “Has the problem” is not observable. “Has hired a sales operations person in the last year” is. The sharper this list, the easier every downstream step becomes.

3. Demographics and firmographics still go in the file, but in their own line labelled “Filters (not the definition).” Keep them separate so you do not mistake the filter for the audience.

4. For every signal or trigger you wrote from your own inference rather than from something you observed or someone told you, log it in assumptions.md (Step: 2, Status: Untested). Step 2.7 will test these.

Step 2.3. Map the decision and the blocker

Duration: 45 minutes

This is the part of Step 2 that most audience work skips, and it is the part that makes the difference downstream. An audience is not just a group with a problem. It is a group facing a decision they have not made yet. The decision is “do something about this or keep living with it.” The reason they have not made it is the blocker. Step 3 will have to beat the blocker, not just the workaround, so name it precisely here.

What to do:

1. In the Step 2 section of outputs.md, under “The decision,” write the actual decision this person faces in their own terms. Not “decide to buy our product.” The product does not exist in their head yet. The decision is closer to “decide whether the way we handle this now is good enough, or whether it is finally worth changing.” State when that decision tends to come up.

2. Under “The blocker,” list what is currently keeping them from making the decision. Work through these candidates and keep the ones that are real for your audience:

   • The status quo is tolerable. The workaround is annoying but survivable, and changing it has its own cost. This is the most common and most underestimated blocker.
   • No clear owner. The person who feels the problem is not the person who can authorise a change.
   • Fear of choosing wrong. They have been burned before, or the choice feels high-stakes and irreversible, so they delay rather than risk a bad call.
   • Competing priorities. The problem is real but not the most urgent thing on the list this quarter.
   • No trigger yet. Nothing has forced the issue. The problem is chronic, not acute, and chronic problems get tolerated indefinitely.

3. For each blocker you keep, note what would have to change for the person to actually decide. That note is a gift to Step 3 and Step 4: it is the thing the message has to make true.

4. The blocker is almost always partly inferred when you first write it. Mark it unconfirmed in the Step 2 section of outputs.md and log it in assumptions.md (Step: 2, Status: Untested). The interviews in Step 2.7 exist largely to confirm or correct this.

Step 2.4. Find the sharpest segment

Duration: 45-60 minutes

A problem is felt unevenly across the people who have it. Some feel it acutely and are ready to act; others feel it mildly and will tolerate it for years. Trying to serve all of them at once produces a value proposition that is true for everyone and compelling to no one. Step 2 forces a choice: which slice of the audience do you aim at first? That slice is your beachhead. Picking it does not throw the rest away; it sequences them. You dominate one slice, use the wins as proof, and expand from there.

Pain also concentrates inside an organisation, not just across organisations. The same company can feel the problem acutely in one team and barely at all elsewhere. The sharpest segment is therefore not just a type of company; it is a type of company plus the specific team or role where the pain lands hardest. Vendors who aim at one function (ops, finance, legal, a particular role) tend to beat generalists, because they can speak that function’s language and solve its exact version of the problem.

What to do:

1. In the Step 2 section of outputs.md, under “Segment intensity map,” lay out the audience along two axes you can fill from your Step 1 consequence range:

   • How acutely do they feel the problem? From “barely notices” to “this is costing them real money or sleep.” Note which team or role inside the organisation feels it most directly, not just which company type.
   • How few blockers are in their way? From “many blockers, will not move soon” to “blocker is already cracking, a decision is close.”

2. Find the segment that sits in the high-intensity, low-blocker corner. These are people who feel the problem and are close to deciding. They are the cheapest to win and the most honest teachers, because they will tell you the truth about the decision and the blocker.

3. Apply the painkiller test. Does this segment treat solving the problem as urgent (a painkiller they are actively reaching for) or as nice-to-have (a vitamin they will get to eventually)? The tells of a painkiller segment are concrete: they are already spending time or money on the workaround, they are actively looking for something better, and a budget either exists or could be found quickly. A segment that reads as a vitamin is the wrong beachhead even if it is large. You will spend the whole sale explaining why the problem matters.

4. Run the segment through the beachhead criteria. A good beachhead is not only high-pain and ready; it is also:

   • Affordable for them. They can pay for a solution at the price the problem justifies.
   • Able to adopt it. They can actually implement and succeed with a solution, so the win is real.
   • Clustered and connected. They talk to each other, go to the same places, and read the same things, so a visible win travels. This is what makes word of mouth and references possible later, and it feeds Step 10 directly.
   • Lucrative enough to matter. Winning the segment is worth the focus.

   Drop a segment that fails these even if the pain is high. A high-pain segment that cannot pay, cannot adopt, or is too scattered for wins to spread is not a beachhead.

5. Score the candidate segments against each other in the matrix below (it is also in the Step 2 section of outputs.md). Use a 1-5 scale in each column. The matrix forces an honest comparison; it does not replace judgment. A 1 on “can afford” or “can adopt” usually disqualifies a segment no matter how high the total, because a win there is either unaffordable or will not stick.

   Candidate segment	Pain (1-5)	Readiness, few blockers (1-5)	Can afford (1-5)	Can adopt (1-5)	Clustered and connected (1-5)	Lucrative (1-5)	Total

6. Write the chosen segment as a single named slice in the “Sharpest segment (beachhead)” field, including the team or role where the pain concentrates. Be specific enough that it passes the “list of ten” test on its own, not just the broader audience. “Mid-size manufacturers” is a category. “Operations planners at mid-size manufacturers running production across more than three systems” is a beachhead.

7. Write one sentence on why this segment feels the problem more than the rest. If you cannot, you have not actually found a sharp segment; you have just relabelled the whole audience. Try again.

8. Note which segments you are deliberately deferring (not abandoning) and why. Dominating one slice and expanding from it beats spreading thin across five. This keeps the scope honest and stops the audience from quietly broadening back out in later steps.

9. Whether a segment really feels the problem more acutely, and treats it as a painkiller, is testable, not obvious. Log the choice as an assumption in assumptions.md (Step: 2, Status: Untested) until interviews confirm it.

Step 2.5. Size the segment

Duration: 45-60 minutes

A segment can be sharp, high-pain, and ready to move, and still be too small to sustain the business, or so large that the “segment” is secretly the whole market again. Neither is visible until you put numbers on it. This step is not an investor slide. It is a go/no-go check on the beachhead you just chose, done in an hour with napkin math, before every step below inherits the choice.

Size bottom-up, not top-down. Top-down (“the market is 4 billion, we only need 1 percent”) produces big numbers and no information, because it never touches the question of who you can actually reach. Bottom-up starts from the observable signals in Step 2.2: count the accounts that fit, price the problem, and multiply. The count is checkable, the price is anchored on evidence you already have, and the result is a number you can defend to yourself. If you want the classic labels: the whole-market number you are deliberately ignoring is TAM, the segment value below is your SAM, and the obtainable value is your SOM. The labels matter less than the bottom-up method.

What to do:

1. Count the accounts. Take the sharpest segment from Step 2.4 and estimate how many real companies (or people) fit it, using the observable signals as filters. Use sources you could actually build a list from: a LinkedIn Sales Navigator search on the role and firmographic filters, industry association directories, review-site category counts, job postings that carry your signals. Write the method next to the number. A count with no method is a guess wearing a suit.

2. Anchor a plausible annual price. You are not setting pricing here; you need a defensible order of magnitude. Anchor it on the Step 1 evidence: what does the problem cost the segment per year, and what do they already spend on the workaround? A solution can usually charge a fraction of the problem’s cost. Write it as a range.

3. Multiply. Accounts times plausible annual price is what the segment is worth per year if you won all of it.

4. Apply a realistic win share. Nobody wins a whole segment. For a focused vendor with a beachhead this sharp, 10 to 25 percent of the segment over two to three years is ambitious but defensible; more than that is a claim you would have to argue for. Multiply again to get the obtainable value per year.

5. Set the viability threshold before you look at the result, and write it down: what does winning this segment have to be worth per year for the focus to make sense? For a funded company this is usually “enough revenue to prove repeatability and fund the push into the next segment.” For a bootstrapper it may be a target income. The threshold is yours; the discipline is setting it before the math tempts you to move it.

6. Fill the sizing table (it is also in the Step 2 section of outputs.md) and write the verdict:

   Line	Value	Method / source
   Accounts in segment (bottom-up count)
   Plausible annual price (anchored on problem cost)
   Segment value per year (accounts x price)
   Realistic win share
   Obtainable value per year
   Viability threshold (set before the math)
   Verdict

   • Clears the threshold at a defensible win share: the beachhead holds. Proceed to Step 2.6.
   • Clears only at a heroic win share (over 30 percent): the segment is too small on its own. Before rejecting it, check what it opens: a small beachhead that leads naturally into adjacent segments can still be right, because dominance travels. Name the adjacent segment it would open and note the sequence in the scope notes. If it opens nothing, go back to Step 2.4.
   • Cannot clear even at 100 percent: it is a niche, not a beachhead. Go back to Step 2.4 and either pick a different high-intensity slice or widen this one a single notch (one filter looser), then re-run the math. Do not widen more than one notch at a time; that is how the audience quietly becomes everyone again.

7. Sanity-check the count both ways. If the number is huge but you could not name ten accounts, the count is a firmographic fantasy; tighten the filters until the count describes companies you could actually find. If the count is small, that is not automatically a failure: a few hundred nameable accounts at a high price is a workable beachhead, and one you can market to by name.

8. Everything in the table is an assumption: the count method, the price anchor, the win share. Log them in assumptions.md (Step: 2, Status: Untested). The interviews in Step 2.7 give a first correction (does the price anchor survive contact with what people say the problem costs them?), and Step 6’s live funnel eventually tests the count for real.

Example (continuing the follow-up-system case):

A Sales Navigator search on B2B software companies, 11 to 50 employees, with two or more AE roles, returns about 4,000 companies. Tightening to the segment’s signals (Series A raised in the last 18 months, a second or third AE hired recently, no enforced follow-up tooling where detectable) leaves roughly 1,200 to 1,500 reachable accounts. The Step 1 evidence put the cost of dropped follow-up at several lost deals a year, tens of thousands in pipeline, so an annual price of 6,000 to 10,000 is a defensible fraction; take 8,000. Call it 1,400 accounts times 8,000: about 11 million per year of segment value. At a 15 percent win share over three years, the obtainable value is about 1.7 million per year. The threshold, set beforehand: 1 million a year from the beachhead proves repeatability and funds expansion into the Series B segment this beachhead naturally grows into. Verdict: clears, without heroic assumptions. Logged as assumptions: the 1,400 count leans on the AE-hiring signal being detectable, and the 8,000 price is anchored on interview claims about lost pipeline, not on anyone’s stated willingness to pay.

Step 2.6. Map the buying context

Duration: 30-45 minutes

In B2B, the person who has the problem is rarely the only person in the decision, and often not the person who signs. The single most common Step 2 mistake is to define the audience as the person who feels the pain and then discover in Step 6 and Step 8 that someone else controls the outcome. Map the whole group now, while it is cheap.

What to do:

1. Determine the buying motion for your sharpest segment. It decides who Step 5 and Step 6 aim at first.

   • Top-down. The decision starts with leadership. Common in enterprise and strategic purchases.
   • Bottom-up. The decision starts with the end user. Common in product-led growth, developer tools, and self-serve SMB.
   • Consensus. Several stakeholders must agree. Common in mid-market.

   Use your interviews to tell which: when people said “I would need buy-in from…”, who did they name, and who moved first?

2. Map the decision-making unit in the table below (it is also in the Step 2 section of outputs.md). Fill only the rows that are real for your case. In a small or self-serve sale, several rows collapse into one person; write it down anyway, because even a single consumer often has an approver (a partner, a boss, a peer).

   Role	Who (title in your case)	Influence	What they care about	How to reach them
   Champion (has the problem, wants it solved)		High	Solving their pain, looking good internally	Directly; they are the one looking
   Economic buyer (controls the budget)		High	ROI, risk, strategic fit	Through the champion; executive-level proof
   Influencer (shapes the call without owning it)		Medium	Technical fit, integration, effort to switch	Product detail, demos
   End user (uses it daily)		Medium	Usability, workflow fit, time saved	Trial, onboarding
   Blocker (can veto or stall)		High, negative	Security, compliance, budget, status quo	Objection handling, references
   Gatekeeper (controls access)		Low to medium	Process, vendor requirements	Procurement and compliance docs

3. Mark who you actually market to versus who signs. The person you reach (usually the champion) is rarely the one who authorises the spend (usually the economic buyer). When they differ, say so: Step 4 will need to speak to more than one of them, and Step 8 will need to arm the champion to sell internally.

4. Sketch the buying process for the segment, even briefly: what triggers the search, who researches, who evaluates, who makes the final call, the rough timeline, and where it stalls most often. The stall point is usually where a blocker from Step 2.3 lives.

5. Note which rows you have evidence for and which are assumed. The composition of the buying group is an assumption until someone in it confirms it. Log it in assumptions.md (Step: 2, Status: Untested).

Why this matters for downstream steps: Step 5 channels and Step 6 lead generation aim at whoever you can reach, usually the champion. Step 8 sales enablement arms that champion to win over the people they cannot reach for you. If the buying group is wrong here, both of those steps aim at the wrong person and you will not see it until deals stall late.

Step 2.7. Validate with interviews

Duration: 2-5 hours active work, 3-7 days calendar time | Target: 5-10 conversations

Step 1 interviews confirmed the problem is real among the people who have it. Step 2 interviews do not re-test that. They sharpen which slice of those people to aim at: which segment feels the problem most acutely and is closest to acting, the decision they face, what blocks it, and who else is in it. Some of your Step 1 interviewees may be the right people again; the questions are different now, aimed at the person, the decision, and the buying group rather than at the problem itself. Aim for five conversations minimum, ideally drawn from your sharpest segment. Stop when you stop being surprised.

Do not substitute desk research for this. Secondary sources can tell you a segment exists; only a conversation tells you what actually stalls the decision and who really signs.

7a. Recruit from the sharpest segment

Duration: 30-60 minutes to set up, then wait for responses

Where Step 1 recruited anyone with the problem, Step 2 narrows. You want people who fit the sharpest segment from Step 2.4, and you want a mix of buying-group roles if you can get them: a champion who has the problem and, ideally, someone who controls budget.

Who to target:

Use your observable signals list from Step 2.2 as recruitment criteria. The signals were built to be applied from outside, so they double as a filter for who to approach. Prioritise people who show the high-intensity, low-blocker signals.

How to find them:

• Warm network first. Highest conversion, most honest conversations. Ask for introductions to people who match the signals, not just anyone in the industry.
• LinkedIn outreach. Search on the observable signals (role plus the behavior or situation), not just job title. The research framing converts best.
• Existing Step 1 interviewees. Go back to the ones who fit the sharpest segment. They already trust the conversation, and you can go deeper on the decision and the buying group this time.
• Communities and review sites. People articulate about the workaround are usually articulate about the decision they made or avoided.

Outreach message template:

Hi [name], I am researching how [segment described by behavior, not title] decides whether to change how they handle [problem area]. Not a sales call, I am genuinely trying to understand the decision and what gets in the way. Would you have 20 minutes in the next couple of weeks?

Target: 5-10 conversations, weighted toward the sharpest segment, booked across one to two weeks so synthesis stays coherent.

7b. Run the interview

Duration: 20-30 minutes per interview

The goal is to build a picture of the person at the moment of the decision: what situation they are in, what decision they faced or are facing, what stopped or slowed them, and who else was in the room. That picture is what you will use in Step 2.8 to write the audience definition from evidence rather than inference.

Always record with permission. Open the same way every time:

“Thanks for making time. I want to understand how you decide whether to change the way you handle [problem area], and what gets in the way. Your actual experience, not what you think I want to hear. Okay to record?”

Follow the thread the person opens rather than reading this as a script. If they reveal something surprising about the decision or who was involved, go deeper there.

Segment-fit questions (5 minutes), confirming they are who you think

You are checking that this person actually sits in the segment you defined, and noticing the signals that placed them there.

• Walk me through what your role looks like day to day.
• Where does [problem area] show up in that, and how often?
• What changed recently that made this more or less pressing?
• Who do you know with the same setup who would not care about fixing this? What makes them different? (surfaces the anti-audience)

Decision questions (8 minutes), understanding the choice they face

This is the heart of Step 2. You want the real decision, in their words, and its timing.

• Have you ever seriously considered changing how you handle this? What prompted that?
• Walk me through the last time you thought “we should do something about this.” What happened next?
• What would have to be true for you to actually make a change?

Blocker questions (7 minutes), understanding what stalls the decision

You are testing the blocker you mapped in Step 2.3. Listen for the reason behind the reason.

• The last time you considered changing and did not, what stopped you?
• What is the risk, for you, in getting this choice wrong?
• Is this the most pressing thing on your list, or does it keep getting pushed? Why?

Buying-group questions (5 minutes), understanding who else is in it

You are confirming the map from Step 2.6.

• If you decided to change this, who else would need to be involved?
• Who controls the budget for something like this?
• Has anyone ever pushed back on a change like this, or could they? Who, and why?
• Where do you go to look into options when you are considering a change like this?

Closing

• Is there anything about this I did not ask that you think matters?
• Is there someone else, especially someone who decides or signs off, you think I should talk to?
• Can I follow up if I have more questions?

7c. Capture immediately after each interview

Do this within an hour of each conversation, before the next one. For each interview, add an entry to the Step 2 section of captures.md using this structure:

Interview [number]
Interviewee: [role, company size or type, anonymise if needed]
Date:

Segment fit (do they match the sharpest segment? which signals?):

Top 3 quotes (verbatim):
1.
2.
3.

The decision they faced (in their words):

The blocker (what actually stopped or slowed them):

Buying group (who else was involved, who signs, who can veto):

Where they look for options (feeds Step 5):

Anti-audience signal (did they look like a fit but turn out lukewarm or wrong? why?):

Surprises (anything you did not expect):

Any Step 2 assumptions this confirmed or challenged:

The verbatim quotes about the decision and the blocker are the most valuable output. They are the raw material Step 3 and Step 4 will use to write a value proposition and message that actually move this person. Capture them exactly.

7d. Synthesise across interviews

Do this after at least five conversations. Go through all your captures in captures.md and update the Interview synthesis fields in the Step 2 section of outputs.md, covering:

Segment confirmation. Did the people you reached actually match the sharpest segment? If the strongest pain showed up in a different slice than you expected, your beachhead may be wrong. Revise Step 2.4.

Painkiller or vitamin. Read the urgency, not just the agreement. Painkiller tells: they ask how soon they could start, a budget already exists, they followed up with you rather than the other way around, they say some version of “I have been looking for this.” Vitamin tells: polite interest that goes nowhere, “interesting, let us revisit next quarter,” and silence after a good conversation. If the segment keeps reading as a vitamin, the beachhead is wrong no matter how real the problem is. Note which way it leans, with evidence.

The real decision. How did people actually describe the decision they face? Rewrite “The decision” from their language, not yours.

The real blocker. Which blocker came up most? Was it the one you mapped in Step 2.3, or something else? The blocker that surfaces repeatedly is the one Step 3 must beat. Quote it verbatim.

Buying group pattern. Who was consistently in the decision? Did the person with the problem usually control the budget, or was that someone else? Update the buying context map.

Where they look. Where did people say they go for information and options? Save this verbatim. Step 5 reads it directly.

Observable signals, confirmed. Which of your Step 2.2 signals actually predicted who had the problem acutely? Drop the ones that did not. Tighten the list.

Anti-audience signals. Who looked like a fit on paper but was lukewarm or wrong in practice, and what set them apart? The lukewarm interviewees are as informative as the enthusiastic ones. Write the signals that disqualify a prospect, so later steps do not waste effort chasing them.

Surprises. What came up that you did not ask about? These are often the most strategically important findings.

After synthesis, mark the relevant rows in assumptions.md as Validated or Invalidated, and note what the evidence showed.

7e. If you genuinely cannot get interviews

If you cannot reach anyone in the time available, use secondary sources as a temporary substitute, not a permanent one.

• Review sites for the workaround tools, read with an eye for who is reviewing and what decision they describe making.
• Community threads, LinkedIn posts, and subreddits where people in the segment describe the decision and what stopped them.
• Job postings that imply the buying context (a role created to own the problem tells you who the champion is; the reporting line tells you who the economic buyer might be).

Document each source in the Step 2 section of captures.md using the same capture structure, noting that it is secondary. Log the absence of interviews as an assumption in assumptions.md. A Step 2 built entirely on secondary sources is weaker, and that needs to be visible to every step below.

Step 2.8. Write the final audience definition

Duration: 30-45 minutes

You now have a behavior-based description, a decision and a blocker, a chosen segment, a buying-group map, and primary evidence. Write the final version.

A complete audience definition has seven components. Write one or two sentences for each:

1. Segment: The sharp, listable slice you are aiming at first, including the team or role where the pain concentrates.
2. Defining behavior or situation: What they are doing and what triggers the problem. Not demographics.
3. The decision: The choice they face, in their terms.
4. The blocker: What is currently keeping them from making it.
5. Buying context: Who else is in the decision (for B2B), and who you actually market to versus who signs.
6. Observable signals: How you would recognise one from the outside, as a short checklist.
7. Anti-audience: Who looks like a fit but is not, and the signals that disqualify a prospect. This keeps Step 5 and Step 6 from spending on people who will never move.

Keep it as structured fields or fold it into a paragraph plus a checklist. Format does not matter. Precision does.

Template:

[Segment] who [defining behavior or situation] face the decision of [decision]. They have not made it because [blocker]. The person who feels this is usually [champion], but [economic buyer] controls the budget and [potential blocker role] can veto. You can spot one because [observable signals]. They are not the right fit when [anti-audience signals].

Example (filled in, continuing the Step 1 example):

Series A B2B SaaS companies that run a direct sales team without any shared follow-up system, where reps each chase warm prospects in their own way, face the decision of whether to keep relying on individual rep discipline or put a system in place. They have not made it because the current mess is survivable and nobody owns the fix: the VP of Sales feels the lost deals but the founder controls new tooling spend and worries a new system will just be another thing reps ignore. You can spot one because they have recently hired a second or third AE, have no defined follow-up cadence, and the VP of Sales has complained publicly about deals “going dark.” They are not the right fit when there is a single founder-seller (no team to coordinate yet) or when an existing CRM is already enforced, because then the problem is mild and the status quo holds.

What to do:

1. Write the final definition directly in the Step 2 section of outputs.md, in the “Audience definition (final)” field.
2. Read it once as a skeptical colleague who wants to argue with it. Where could you not actually build a list? Where is the blocker still a guess? Tighten those parts.
3. Run the diagnostic from the top of this page one more time. If all six questions now have clear written answers, Step 2 is done.

What you’ve built

After completing the steps above, the Step 2 section of outputs.md should contain:

This is not a deliverable for anyone else. It is a constraint on Step 3.

Assumption sweep

Before moving on, scan the Step 2 section of outputs.md for any field you filled in from reasoning rather than evidence. Common ones at Step 2:

• The observable signals (did you confirm these predict the problem, or assume they do?)
• The blocker (did someone describe what stopped them, or did you infer it?)
• The sharpest segment (did interviews confirm this slice feels it most, or is it your best guess?)
• The sizing inputs (is the account count from a real list source, is the price anchored on evidence of what the problem costs, is the win share more than hope?)
• The buying group (did someone in it confirm the roles, or did you assume the structure?)
• The decision timing (did people tell you when it comes up, or did you imagine it?)

Each unconfirmed field is an assumption. Log it in assumptions.md now if you have not already. A long list here is honest, not a failure. It is the list the next round of evidence has to close.

What this step hands off to Step 3

Before moving on, confirm the Step 2 section of outputs.md is complete. Step 3 opens by reading it. Specifically, Step 3 needs:

• The sharpest segment, defined by behavior, that the value proposition will aim at.
• The decision the segment faces, because the value proposition is the answer to that decision.
• The blocker, because the value proposition has to beat the blocker, not just the named workaround from Step 1. A value proposition that ignores the blocker convinces people who were never going to be stopped by it and misses the ones who are.
• The buying context, because “why choose you over the alternative” has a different answer for the champion than for the economic buyer.

The sharpness of your Step 2 audience sets the ceiling on how compelling your Step 3 value proposition can be. A value proposition written for a vague audience is vague by necessity. If Step 3 keeps coming out generic, the cause is usually here: the segment is too broad or the decision is not specific enough. Come back and narrow.

The other thing Step 2 hands off: where the audience looks for options, captured in Step 2.7. Step 5 reads that directly when it chooses channels.

The segment sizing from Step 2.5 also travels: Step 6 sanity-checks its funnel volume against the segment count (the funnel cannot produce more leads than the segment contains), and the price anchor is an early input to the value conversation in Step 3.

Common failure modes

The audience is a demographic filter in disguise. “Marketing leaders at 200 to 500 person SaaS companies” describes who they are, not why they have the problem or whether they feel it. If your definition would not change at all when the problem changes, it is a filter, not an audience. Add the behavior and the situation.

The audience is everyone who has the problem. No sharpest segment was chosen, so Step 3 has to write a value proposition true for all of them, which makes it compelling to none. Pick the beachhead. Defer the rest in writing.

The beachhead is a puddle. High pain, ready to buy, and forty accounts in the world at a price that cannot sustain the business. Nobody sized the segment, and the funnel runs dry in Step 6 where the fix is expensive. The reverse failure is just as common: a huge top-down number justified keeping the segment vague. Size bottom-up in Step 2.5 and let the verdict bite.

The user and the buyer got conflated. You defined the audience as the person who feels the pain and assumed they can act on it. In B2B they often cannot. Map the buying group and name who actually signs.

The blocker was ignored. You assumed that because someone has the problem and knows it, they are ready to act. Most are not; the status quo is tolerable and changing has its own cost. If Step 2 has no named blocker, Step 3 will be written as if the only competition is the workaround, when the real competition is inertia.

The audience drifted from Step 1. You started from Step 1’s “Who” and quietly broadened it back out to make the audience feel bigger. Notice the drift and pull it back. A bigger audience on paper is not a sharper one.

You are aiming at vitamins. The segment agrees the problem is real but treats solving it as nice-to-have, so nothing moves. The tell shows up later as a dragging sales cycle, constant re-explaining of why you matter, and warm conversations that go silent. If that is happening, the cause is usually here: you picked a segment that feels the problem mildly. Move to the segment that treats it as a painkiller.

“We know our audience.” Said with confidence, with no written definition that passes the “list of ten” test. If you cannot point to the definition in outputs.md, it is not done.

Sources

• Crossing the Chasm by Geoffrey A. Moore. The source for choosing a single beachhead segment instead of pursuing the whole market at once, and for target customer characterization built around situation and behavior rather than demographics. Step 2.4 leans on this directly.
• Disciplined Entrepreneurship by Bill Aulet. The source for bottom-up market sizing of a single beachhead: counting the actual accounts that fit, pricing from the value delivered, and judging whether the beachhead can sustain the business before committing to it. Behind the napkin math in Step 2.5.
• Obviously Awesome by April Dunford. On defining best-fit customers by the characteristics that make them care, and on why a sharp audience is the precondition for positioning. Useful background for Steps 2 and 8 and the handoff to Step 3.
• The JOLT Effect by Matthew Dixon and Ted McKenna. On customer indecision: why people who acknowledge a problem still fail to act, and what actually stalls a decision. The foundation for the blocker work in Step 2.3.
• The Mom Test by Rob Fitzpatrick. The interview method reused in Step 2.7, here aimed at the decision and the people in it rather than the problem. Covers how to get honest answers about what people actually did, not what they say they would do.
• The New Strategic Selling by Robert B. Miller and Stephen E. Heiman. The source for mapping the buying group by role (economic buyer, user, technical or influencer, coach or champion) rather than by title. The foundation for the decision-making unit table in Step 2.6.