How to Find Your Startup ICP
Most founders sell to everyone and learn from no one. Here's how to define your startup ICP so sales, product, and positioning get sharper.
Published , updated · 9 min read
Most early-stage founders do not lose because they picked the wrong market. They lose because they tried to learn from too many markets at once.
One week they sell to agencies. The next week they talk to SaaS founders. Then a healthcare operator says something interesting, so they adjust the deck. Then an investor suggests enterprise. The product gets broader, the pitch gets blurrier, and every customer conversation teaches a slightly different lesson.
That is what happens when you do not have an ICP.
ICP stands for ideal customer profile, but that phrase makes it sound softer than it is. Your ICP is not a persona poster with a stock photo and a name like "Operations Olivia." It is the narrow customer segment where your problem, product, sales motion, and timing have the best chance of compounding.
Your ICP is the answer to one painful question: if you could only learn from one kind of customer for the next 90 days, who should it be?
Getting this right makes almost every other founder problem easier: customer interviews, cold email, demos, pricing, onboarding, roadmap choices, and even your pitch deck. Getting it wrong makes everything look noisy.
Why Most Founders Define ICP Too Late
The common advice is to keep the market broad until you see pull. That sounds reasonable, but it often creates the opposite of learning.
When every conversation comes from a different customer type, you cannot tell whether objections are product problems, segment problems, positioning problems, or timing problems. One buyer says the price is high. Another says the feature set is too narrow. A third says they would use it if their team were larger. A fourth loves it but has no budget.
Those are not four clean data points. They are four different markets talking over each other.
An ICP gives you a controlled learning environment. You are not claiming this is the only customer you will ever serve. You are choosing the customer whose signal deserves priority right now.
Start With Behavior, Not Demographics
Bad ICP work starts with labels:
- B2B SaaS companies
- Small businesses
- Marketing teams
- Founders
- Agencies
Those labels are too broad to guide anything. A 3-person SaaS company selling to consumers behaves nothing like a 75-person SaaS company selling to banks. A founder with urgent pipeline pain behaves nothing like a founder who is casually exploring tools.
Start with behavior instead:
- They built a spreadsheet workaround in the last 30 days.
- They already pay for three tools that only partly solve the problem.
- They personally feel the pain every week.
- They have an upcoming deadline that makes the pain urgent.
- They asked for a demo without needing to be convinced the category matters.
Behavior tells you where energy already exists. Demographics only tell you where a person sits in a database.
The Six Scores That Matter
Make a list of the customer types you have touched so far. Then score each one from 1 to 5 across six dimensions.
| Score | Question |
|---|---|
| Pain severity | How expensive, frequent, or embarrassing is the problem? |
| Urgency | Why does this need to be solved now instead of someday? |
| Budget clarity | Who can pay, and is this already a budget line? |
| Access | Can you reach these people repeatedly without heroic effort? |
| Switch willingness | Are they actively unhappy with the current workaround? |
| Retention signal | If they buy, is the problem recurring enough that they stay? |
The best ICP is rarely the biggest market on paper. It is the segment with the strongest combined score.
This is where founders often confuse market sizing with go-to-market sequencing. TAM tells you whether a market can become large. ICP tells you where to start learning and selling.
Watch for the "Easy Conversation" Trap
Some segments are pleasant to interview but terrible to sell.
They reply quickly. They give thoughtful feedback. They enjoy talking to founders. They say the product is interesting. Then nothing happens. No urgency, no budget, no champion, no next step.
That is not an ICP. That is a feedback audience.
A real ICP does not just give you insight. It creates movement. After a good conversation, there is a next step: another stakeholder, a trial, a pilot, a document request, a paid test, a technical review, a referral, or a clear no.
The founder mistake is treating polite intelligence as buying intent. Your ICP should be defined around customers who move, not customers who merely talk.
Write the ICP as a Testable Hypothesis
Do not write:
Our ICP is early-stage startups.
Write:
Our ICP is seed-stage B2B SaaS founders with 3 to 15 employees who are preparing for their next fundraise, currently manage investor updates and operating metrics across spreadsheets and docs, and feel weekly pain from scattered company context.
That version gives you something to test. It includes:
- Company stage
- Business model
- Role
- Trigger event
- Current workaround
- Pain pattern
- Reason to act now
If you cannot write the current workaround, you probably do not understand the segment yet. If you cannot write the trigger, you probably do not know why they will buy now.
The One-Week ICP Sprint
You can get useful ICP signal in a week if you keep the scope tight.
Day one: list every plausible segment and score them across the six dimensions.
Day two: pick the top segment and write a one-paragraph ICP hypothesis.
Day three: write a focused message using that segment's language. No generic pitch. Name the pain, the trigger, and the current workaround.
Days four and five: run 8 to 12 conversations or sales calls with only that segment.
Day six: review the evidence. Did they recognize the pain quickly? Did they describe the workaround without prompting? Did they ask buying questions? Did they accept a next step?
Day seven: decide whether to commit for another 30 days, narrow the segment, or move to the next candidate.
This is not enough to prove a market. It is enough to stop guessing blindly.
How ICP Changes Your Product
A sharp ICP removes false roadmap pressure.
Without an ICP, every request sounds equally valid. The agency wants client reporting. The founder wants investor updates. The enterprise buyer wants SSO. The solo operator wants a cheaper plan. The product becomes a pile of reasonable requests from incompatible customers.
With an ICP, you can ask a better question:
Does this request make the product more valuable for the customer we are choosing to win right now?
That does not mean ignoring everyone else forever. It means your roadmap stops lying to you. You can say no because the no is attached to a strategy, not to founder mood.
How ICP Changes Sales
A sharp ICP also makes founder-led sales less awkward.
Your cold email improves because the first line names a real situation. Your discovery calls improve because you can ask specific questions. Your demo improves because you do not tour every feature. Your follow-up improves because it refers to a known trigger.
Most founders think sales gets easier when the product has more features. In the early stage, sales usually gets easier when the customer definition gets narrower.
The Signal That Your ICP Is Working
You know the ICP is getting sharper when five things start happening:
- Prospects describe the problem in similar language.
- The same objections repeat.
- The same current workaround appears.
- Your demo gets shorter because you know what to show.
- Your content, emails, landing page, and product roadmap start using the same words.
This is the compounding effect. Every customer conversation improves the next one because the inputs are comparable. That is how you move from scattered learning to a real go-to-market motion.
What This Looks Like in 1tab.ai
1tab.ai helps founders turn ICP work into a connected operating system: customer interviews, personas, market research, CRM notes, product decisions, and pitch deck positioning live in one workspace, so the ICP is not trapped in a doc nobody opens.