How to Define Your ICP Using Real Data Signals
You know what most ICPs look like? "Mid-market SaaS companies with 50–200 employees who want to grow revenue."
Riveting stuff. Truly. That description fits about forty thousand companies, and it tells your sales team absolutely nothing useful.
Here's the uncomfortable truth: most companies don't have an ICP. They have a description someone wrote on a Friday afternoon that's never been stress-tested, never been updated, and lives in a Google Doc no one reads. That's not an ICP — it's a guess with good formatting.
A real ICP is built from data signals. Not vibes. Not "we feel like our best customers are in fintech." Actual signals from your actual customers that tell you exactly who closes fastest, stays longest, and shouts your name from the rooftops.
Here's how to build one.
Start With Your Best Customers — Not Your Most Recent Ones
Pull your top 10–15 customers. Not the ones who signed up last month. The ones who:
- Closed quickly
- Churned the least (or not at all)
- Expanded their contract over time
- Referred other customers to you
This is your starting point. These people are showing you something. Your job is to figure out what.
Extract the Firmographic Signals
Once you have your list, go company by company and document the basics:
- Industry / vertical
- Company size (headcount and revenue, separately — they tell different stories)
- Geography
- Funding stage (bootstrapped vs. Series A vs. enterprise changes the buying conversation completely)
- Tech stack (especially if your product integrates with something)
You're looking for patterns. If eight out of your top ten customers are Series B SaaS companies with 30–80 employees using HubSpot, that's not a coincidence — that's your ICP waving at you.
Go Deeper: Behavioural and Situational Signals
Firmographics tell you who they are. Behavioural signals tell you why they bought. This is where most companies stop short and pay for it later.
Go back to your CRM and your customer conversations and ask:
- What triggered the purchase? What was happening in their business when they came to you? Rapid hiring? A failed implementation? A new VP who wanted to change everything?
- What were they using before? Switching from a competitor tells you one story. Switching from spreadsheets tells you a completely different one.
- How long was the sales cycle? Fast closers are gold. If certain customer types close in two weeks and others take six months, that's a signal worth building into your targeting.
- Who was in the room? What title actually signed the deal? Not who you pitched first — who said yes.
These situational patterns are the difference between an ICP that sounds good and one that actually works.
Talk to the Customers Themselves
Yes, this means picking up the phone. Or getting in a room. Radical concept.
Ask your best five customers three questions:
- What was happening in your business right before you started looking for a solution like ours?
- What almost stopped you from buying?
- What would you tell someone in your position about why we were the right call?
The language they use in those answers? That's your messaging. The trigger they describe? That's your targeting signal. The objection they almost had? That's what your sales team needs to handle in every first call.
This is why the B2B growth strategy the complete playbook treats ICP definition as infrastructure, not an afterthought — because everything downstream of it, from outreach to content to how you run a discovery call, depends on getting this right first.
Build the Document. Actually Build It.
A new hire should be able to read your ICP document on day one and immediately know:
- Who to target
- Who to deprioritise
- What triggers to look for
- What titles to reach
- When to make an exception (and what an exception even looks like)
If that document doesn't exist, you don't have an ICP. You have an intention.
The document should include your firmographic profile, your behavioural triggers, your "green flags" (signals that someone's a great fit), and your "red flags" (signals that it'll be a painful deal). One page. No jargon. Usable on day one.
Revisit It. Seriously.
Your ICP is not a set-and-forget exercise. Markets shift. Your product evolves. The customers you're winning in month six look different from the ones you closed in month one.
Build a habit of reviewing it every quarter. Pull the same data. Look for drift. If your best customers are starting to cluster somewhere new, follow them.
Most companies skip the data work and pay for it in long sales cycles, high churn, and a sales team that's targeting the wrong people with the right message.
Do the unglamorous work. Pull the list. Extract the signals. Build the document.
That's what a real ICP looks like.
