Great customer discovery isn't about doing "more interviews." It's about interviewing the right people and being honest about what segment each insight came from. This guide shows a lightweight way to target your ICP while you're still exploring: define candidate segments, recruit intentionally, track simple attributes, and synthesize what each segment wants so you can choose where to focus. Getting this right is existential: CB Insights' analysis of failed startups found 43% cited poor product-market fit — building for the wrong customer — as a reason they shut down.
Customer discovery fails quietly when you talk to the wrong mix of people
Most founders can do 10–15 conversations. The problem is what those conversations represent.
If you interview:
- a buyer and a daily user
- a startup and an enterprise team
- someone in the core workflow and someone adjacent
…then you'll hear different wants, different constraints, and different language. That's normal.
What's not normal is treating all of that as one blended dataset and calling the result "validation."
Early-stage discovery gets dramatically easier when you do two things:
- target the right people for the question you're asking, and
- track enough context to interpret what you hear later.
Step 1: Start with a "candidate ICP list" (2–4, not 12)
When founders say "I don't know my ICP yet," that's fine. You don't need certainty to start — just a few reasonable candidates.
Pick 2–4 candidate segments you want to learn about this month.
Examples:
- Founders at pre-seed who do customer interviews weekly
- Sales leaders at 10–50 person startups doing outbound
- Ops managers at 50–200 person teams managing handoffs
- Founding PMs building early discovery loops
You're not committing. You're creating a controlled experiment: Which segment gives the clearest pain + urgency + willingness to change?
Step 2: Decide what "counts" for each segment
Before you book a bunch of calls, write 2–3 "qualifying" rules for each candidate segment.
Example:
- Must be a founder or founding PM
- Team size 1–20
- Actively doing discovery interviews or trying to land first customers
This avoids "interesting conversations" that aren't actually relevant to the outcome you want.
Step 3: Recruit intentionally (so you don't just interview whoever replies)
A simple coverage goal keeps you honest.
For your first 15 interviews, try something like:
- 6 in Segment A
- 6 in Segment B
- 3 "adjacent" (useful contrast, but label them as such)
Recruiting sources that work well early:
- warm intros
- LinkedIn search
- event attendee lists (Luma is great for SF)
- founder communities (Slack groups, Indie Hackers, etc.)
The point isn't where you find people. The point is: you know which segment they belong to.
Step 4: Use a script that reveals reality (not opinions)
If you're targeting ICP, your questions should do two jobs:
- uncover pain and workflow truth
- reveal whether they truly match the segment
Here's a tight structure:
A) Segment confirmation (1–2 minutes)
- "What's your role in this workflow?"
- "What's your team size / stage?"
- "How often do you run into this problem?"
B) The "last time" question (signal gold)
- "Tell me about the last time this happened."
- "What triggered it?"
- "What did you do next?"
C) Current solution + switching friction
- "How do you handle it today?"
- "What do you like about that?"
- "What do you hate about it?"
- "What would make you change?"
D) Close (keep it founder-friendly)
- "Who else should I talk to in this segment?"
- "If I build a rough prototype around this, can I show you?"
Notice what's missing: "Would you use my product?" Early-stage discovery should be about behavior, not politeness.
Step 5: Track a few attributes so your insights stay interpretable
You don't need heavy research ops. You just need enough context so you can later answer: "What's true for who?"
A great minimal set:
Person
- role: user / buyer / manager / founder / PM
Company
- stage or size (pick one)
Optional if relevant
- tool stack/workaround
- workflow maturity
- budget sensitivity
This is the difference between "we heard X" and "Segment A consistently wants X."
Step 6: Synthesize by segment, not as one blended bucket
After 10–15 interviews, don't ask: "What are the themes?"
Ask: "What are the themes for each segment?"
A simple format:
- Segment A: top 3 recurring pains + language + what they want
- Segment B: top 3 recurring pains + constraints + what they want
- Adjacent: interesting but not core
This turns discovery into an ICP decision.
How you know you found an ICP worth focusing on
A segment is usually a good "first ICP" when:
- pain is frequent and specific
- they have urgency now (not "someday")
- they can describe a real workaround
- the buyer/user path is clear enough
- they want a solution that fits your build capacity
You don't need everyone to agree. You need one segment with repeatable clarity.
Where Intervool fits
Intervool is built for exactly this: discovery plus ICP targeting without heavy overhead.
It helps you:
- track targets and whatever attributes matter for your market
- keep interview notes connected to the segment they came from
- synthesize takeaways into themes by segment
- avoid blending audiences and losing the "for who" context
If you're doing discovery right now, Intervool offers a 30-day free trial for early teams.
