How to Validate Your ICP in 2026: A 5-Step Process You Can Run in Two Weeks
Write a specific ICP hypothesis, check public evidence, run 10 to 15 problem calls, ask for a commitment, then confirm with a small outbound test.
To validate your ICP, write a specific hypothesis (role, company size, stage, trigger), check public evidence that the trigger exists, and run 10 to 15 problem interviews without pitching. Then ask the people with real pain for a commitment. Finish with a small outbound test against a second segment. If one segment says yes more often, it's validated.
This is the hands-on route: calls, commitments and a live test. If you have several candidate segments and want to narrow them first, ICP validation with AI shows how to rank them with RightAudience before you book a single call.
Rank your candidate segments with RightAudience →
Who this is for
You've written down who you think your buyer is and you're about to spend on outbound, ads or a hire built around that guess. You want to know it holds up first, using conversations and a small test.
Definition vs validation
Defining your ICP means writing down who you think buys. Validating it means getting evidence they do. The gap matters because a segment can have the problem you solve and still not buy: the pain isn't urgent, nobody owns the budget, or the workaround is good enough. Validation checks for urgency, budget and intent, not just the problem.
The 5-step process
Step 1: Write a hypothesis you can prove wrong (day 1)
"SMB founders" can't be wrong, so it can't be validated. Write it like this:
B2B SaaS founders, 1 to 10 people, pre-seed or seed, who are running their own outbound and have no SDR yet.
Include four things: role, company size, stage and the trigger that makes the problem urgent right now. Write a second segment the same way. You'll test them against each other.
Step 2: Check the public evidence (days 1 to 2)
Before calling anyone, check the trigger exists at scale. Can you find 100 companies that match? Are they posting job ads, writing posts or asking questions in communities that show the problem? Are they already paying for a workaround? If you can't find 100, the segment may be too small or too hard to reach.
Step 3: Run 10 to 15 problem interviews per segment (days 2 to 9)
Talk about the problem, not your product. Ask:
- How do you handle this today?
- What does it cost you in time or money?
- What have you tried? Did you pay for any of it?
- Who decides whether to spend on fixing it?
- What would make this a priority this quarter?
Score each call 1 to 3 on urgency, budget authority and prior spend. How to run customer discovery interviews has a fuller question list.
Step 4: Ask for a commitment (during the calls)
At the end of every call where the pain is real, ask for something that costs them. A paid pilot, a prepayment for early access, a letter of intent with a price, or an introduction to the budget owner. Count the yeses. "Sounds great, keep me posted" doesn't count.
Step 5: Run a small outbound test (days 8 to 14)
Pick 50 to 100 contacts from each segment and send the same email, changing only the details that name the segment. Compare reply rates and how positive the replies are. A segment that replies clearly more often, and more positively, with the same message has stronger intent. How to test a cold email before sending covers setting up a clean test.
How to score and decide
Put your results in one table.
| Segment | Avg urgency (1 to 3) | Avg budget authority (1 to 3) | Avg prior spend (1 to 3) | Commitments | Outbound reply rate |
|---|---|---|---|---|---|
| Segment A | |||||
| Segment B |
Pick the segment that leads on commitments and outbound replies. Interview scores explain why; commitments and replies are the evidence. If neither segment produced a commitment, don't pick the better of two weak results. Rewrite the hypothesis and run another round.
Then commit. Running two segments in parallel splits your messaging, your channels and your attention. Keep the runner-up as the next segment to test once the first is working. For a reusable template, use the ICP scoring matrix.
Symptom, check, fix
| Symptom | Check | Fix |
|---|---|---|
| Every interview sounds different | Is the hypothesis specific on all four parts? | Narrow role, size, stage or trigger and rerun |
| Pain is real but nobody commits | Does the person you're talking to own budget? | Interview the budget owner; ask for an intro |
| Interviews go well, outbound gets no replies | Does the email name the same pain in their words? | Rewrite using phrases from the calls |
| You can't find 100 matching companies | Is the segment too narrow or invisible? | Widen one dimension, or pick a segment you can reach |
| Both segments look the same | Are the segments actually different? | Change the trigger or stage, not just the industry |
A worked example
For example, a founder building a contract-review tool tests two segments: in-house legal teams at 100 to 500 person companies, and solo lawyers. After 12 calls each, in-house teams score higher on urgency and prior spend, and two agree to a paid pilot. Solo lawyers like the idea, but none commit. The outbound test agrees: in-house teams reply more often to the same email.
The founder commits to in-house legal teams, rewrites the homepage for them and parks solo lawyers as the next segment. The numbers are illustrative.
Next step
Write your two segment hypotheses with role, size, stage and trigger today, and book the first five problem interviews for each. If you have more than two candidates, rank them first with RightAudience using the AI route.
Narrow your segments with RightAudience →
Related: ICP validation with AI - How to find your ICP with no customers - How to choose between two target audiences