Most ICP documents are firmographic filters dressed up as strategy.
"We target mid-market B2B tech companies with 200-500 employees, $20M-$100M in revenue, in North America." That's a database filter. It's not an ICP. The difference matters because a database filter tells you who to spray your marketing at. A real ICP tells you who will actually buy — and why they'll buy from you specifically.
Start with your closed-won data
Every ICP development process should start with the same question: who are your best customers and what do they have in common? Not your biggest customers — your best customers. The ones who renew, expand, refer you, and don't drain your CS team.
Pull a list of your top 20-30 customers by lifetime value and retention rate. Then go deeper than firmographics. Talk to your sales team about what was happening inside those companies when they bought. You're looking for patterns in three categories: situational triggers, organizational signals, and technical or operational fit.
The four ICP dimensions
Build your ICP across four dimensions — not two. Firmographic: industry, company size, geography, growth stage. Technographic: what tools they use and what that signals about their sophistication. Behavioral: how do they research, how long is their buying cycle, who owns the decision? Situational: what specific circumstances make them a buyer right now?
The situational dimension is the one most teams miss. An ICP isn't static — a company can be a perfect firmographic fit and not be a buyer today, and then become your ideal buyer the moment something changes inside them.
Tiers aren't optional
One ICP profile is almost never enough. Build tiers based on how closely different buyer segments match your strongest signals. Tier 1 is your highest-fit, highest-value buyer — the one you build core messaging around. Tier 2 is close but with tradeoffs. Tier 3 is reachable with your current motion but not where you want your best ABM resources.
The living document problem
ICPs go stale. The market changes, your product evolves, you win in segments you didn't expect. Build in a quarterly review — a 60-minute conversation with sales leadership asking "is this still true?" is enough. The companies that execute ABM most effectively treat their ICP as a living hypothesis they're constantly testing, not a document that gets filed away.
The shortcut that isn't: building your ICP from first principles rather than research produces a cleaner-looking document that's almost always wrong in ways only visible six months into a campaign. The research is the work. Everything else is formatting.
— Erik R. Miller