How to Build Your ICP From CRM Data in 6 Steps

Your best customers are sitting in your CRM right now. Most teams never look at them.

Instead, they guess. "We target mid-market SaaS companies in the US." Then they wonder why 70% of demos don't convert.

The smarter play: reverse-engineer who you actually win against, then hunt for more of them. Here's how.

Why ICPs Built From Real Data Convert Better

A data-backed ICP is narrow and ruthless. It kills time spent on accounts that don't fit. It focuses your entire GTM machine on the 15-20% of the market that actually buys from you.

Teams without a real ICP: - Spend 40% of sales time on accounts that never close - Run campaigns to the wrong buyer personas - Waste ad spend on companies that look good but don't have budget - Spin their wheels on accounts with multi-year sales cycles

Teams with a real ICP: - Know the company size, industry, and revenue that converts - Know which buyer titles actually engage - Know the tech stack that signals buying readiness - Cut deal cycles by 3-4 months

The difference is time and precision.

Step 1: Pull Your Win Data

Open your CRM. Filter for deals closed-won in the last 18-24 months. You want a big enough sample (ideally 30+ deals minimum) to spot patterns without outliers distorting the picture.

Export: - Company name - Annual revenue - Number of employees - Industry - Geography - Tech stack (what tools they use) - Decision maker titles that participated - Sales cycle length - Deal size (ARR or contract value)

If your CRM doesn't track all of this, start tracking it now for future deals. For today, work with what you have.

Step 2: Segment Wins by Deal Type

Not all wins are equal. A 50-person startup with a $10K ARR deal has a different profile than a 2,000-person enterprise with $200K ARR.

Bucket your wins: - Enterprise: Deals $100K+ ARR, 500+ employees - Mid-market: Deals $30-100K ARR, 50-500 employees - SMB: Deals under $30K ARR, under 50 employees - Bottom-feeders: Deals you didn't want (or shouldn't have won)

Focus on the segment that's most profitable for you (highest ACV, shortest churn, lowest CAC). That's your ICP anchor.

Step 3: Find Common Company Attributes

Grab your win segment. Now ask:

• What industries are over-represented? (If 60% of wins are in fintech, that's a pattern.)
• What's the typical company size (revenue + headcount)?
• What regions do they operate in?
• How many of your wins are using your competitor's tools?

Look for overlap. If you won 25 fintech companies between 20-200 employees, predominantly in the US and EU, that's the core of your ICP. If wins are scattered everywhere, you don't have an ICP yet. You have a product that works for anyone willing to pay.

Step 4: Identify Buyer Personas Within Your ICP

Deals are closed by people, not companies. Find who actually cares.

For each win, note: - Title of the economic buyer (who controls budget?) - Title of the user/champion (who fights internally for your solution?) - Title of the blocker (who else needs to agree?)

If 80% of your wins involved a director of marketing or senior marketing manager as the champion, that's your buyer persona. If the blocker is always compliance, you know to involve legal early.

Most teams miss this step. They define the company but not the actual person making the call. Then they wonder why they can't get past the first meeting.

Step 5: Calculate Your Average Winning Profile

This is the ICP in numbers:

• Company Revenue: $10-100M (example)
• Headcount: 50-500 (example)
• Industry: SaaS, fintech, martech (example)
• Geography: North America, Western Europe (example)
• Key Buyer Titles: Director of Demand Gen, VP Marketing
• Average Deal Size: $45K ARR
• Average Sales Cycle: 4-6 months
• Average CAC: $12K
• Average Churn: 8% annually

This isn't a straitjacket. It's a compass. A company doing $500M in revenue with the right buyer and urgency might still be worth a conversation.

Step 6: Test and Refine

Your first ICP is not your final one. Test it:

• Run an account-based campaign targeting 50 companies that match your ICP
• See which accounts you actually land meetings with
• Look at which conversations turn into pipeline
• Compare your pipeline profile to your ICP definition

Refine every 90 days. If you're finding that your best deals are coming from a smaller subset (say, fintech companies under 200 people instead of 50-500), update your ICP.

The teams that win with ABM are the ones that actually iterate on this.

Why Most Teams Get This Wrong

Common ICP mistakes:

• Too broad: "We sell to any mid-market SaaS company." That's not an ICP. That's a market.
• Based on your feature set: "We sell to companies that need workflow automation." No. You sell to companies with a specific problem AND budget AND urgency.
• Ignoring sales cycle: A 12-month deal to an enterprise looks different than a 4-month deal to mid-market. Both are valid ICPs. They're just different bets.
• Not accounting for CAC: If your ICP company size requires a $150K sales investment but you only close half, your math breaks.

Your ICP Is Your Anchor

Once you have a real ICP, everything else cascades. Your ABM campaigns target those accounts. Your outreach templates hit those buyer personas. Your ad spend focuses on those industries and geographies. Your sales process gets refined around their sales cycle.

Without it, you're guessing. With it, you're hunting.

Build it this week. You'll recognize the companies you actually win against. Then hunt for 100 more just like them.