How to identify your bullseye customer in one day

Executive overview

Most founders build for everyone and stall. A bullseye customer is the specific subset of your target market most likely to adopt first — far narrower than a typical ICP.

Michael Margolis compressed 30 years of UX research into a one-day sprint: five bullseye customers, three prototypes, whole team watching live. No report. Immediate alignment.

Failing to identify a specific initial customer is the most common reason early-stage products get built for nobody.

What a bullseye customer is

• The narrowest viable slice of a target market — who will adopt first, not who will eventually buy
• Amazon started with books; Facebook started with college profiles — focus precedes scale
• Narrows the team's roadmap, filters incoming feedback, and aligns everyone on what to build next
• Distinct from ICP: the bullseye is a research instrument to discover a real ICP, not a sales definition

The sprint formula: five and three in one

• Five bullseye customers, three prototypes, all five interviews in one day
• The whole team watches live — no written report needed afterwards
• Clumping interviews makes patterns obvious; spreading them across weeks loses the signal
• Five interviews hits data saturation — the point where everyone starts hearing the same things

Step 1: define your key questions

• Run a 45-minute team meeting before recruiting or prototyping anything
• Ask: what keeps you up at night? What would have to be true for this to succeed?
• Surface nagging internal debates and unresolved assumptions about the customer and product
• These questions drive all downstream decisions — who to recruit, what to test, what to measure

Step 2: define the bullseye customer

Three categories of recruiting criteria:

• Inclusion criteria — who you want (sector, role, usage patterns, life situation)
• Exclusion criteria — who to set aside: domain experts, locked-in competitor users, people with disqualifying experiences
• Triggers — events that make someone ready now (new baby, new CSO, recent failure with current solution)

Aim for roughly seven concrete, measurable attributes. Push to "comically narrow." If you can't find these people during recruiting, treat that as a signal about the business, not just the research.

Step 3: recruit participants

• Translate criteria into a screener that does not telegraph the right answers; use open-ended questions
• Use userinterviews.com or Respondent; typically 3–4 days to fill five slots
• Pay properly: $125/hour for most consumer sessions; match professional market rates when needed
• Track NDA response rate as a proxy for show-up reliability; swap unresponsive participants early
• If the panel cannot surface your bullseye, question whether those people exist at the scale you need

Step 4: build three prototypes

• Three distinct value-prop "recipes" — not three iterations of one idea
• Flat PDFs or Figma exports only; no functionality required
• Primarily a writing exercise: articulate each value proposition clearly and bluntly, no marketing speak
• Make prototypes visually distinct (colour, layout) so observers can track which one is being discussed
• Use competitor products as free prototypes — most teams underuse this
• Proofread carefully; errors undermine credibility and derail the conversation

Step 5: write the interview guide

Each one-hour session splits into two halves:

• Discovery (first half): past behaviours, experiences, attitudes — what worked, what failed, and why
• Prototype comparison (second half): present each prototype; gather reactions; prompt comparison across all three

Look for the best pieces across prototypes, not a winner. Past behaviour is far more reliable than stated future intent — weight it accordingly.

Interview technique:

• Start with a big smile even on audio — it changes your voice and sets the tone
• Get the person smiling back in the first few minutes before going deep
• Enter "listener mode": extreme curiosity, no pitching, no defending
• Practice the mode switch; it is a different character from founder-pitching mode

Step 6: run the watch party

• Conduct interviews one-on-one over Zoom; live-stream to the team watching separately
• Team takes manual notes in a shared doc — AI note-takers cause people to disengage
• Rotate note-taking assignments per interview; run a Slack back-channel for observer chatter
• A design partner monitors the back-channel and passes urgent follow-up questions to the interviewer
• Debrief after each interview using a spreadsheet: key questions as rows, participants as columns
• Before the day starts, record each team member's specific prediction of what will happen
• End the day with an independent big-takeaways form per person; decider synthesises patterns aloud
• Compare end-of-day findings against pre-day predictions to expose hindsight bias and quantify learning

Common pitfalls

• Mushy recruiting: a bullseye definition that bleeds too wide produces uninterpretable results
• Including known contacts: familiar customers skew results; recruit strangers via panels
• Confirmation bias: build in peer permission to call it out during the watch party
• Over-weighting stated intent: "I'd definitely use this" is weak signal; lean-in energy and "can I sign up?" is strong signal
• Curse of knowledge: expert teams overestimate how much customers know, how acutely they feel the problem, and how ready they are to pay