Framework

ICP Definition Framework

A step-by-step framework for defining a B2B ICP from deal history and buying-readiness signals, with a worked example and the mistakes that quietly undo it.

Most ICP documents are written once, in a workshop, from assumptions about who the “ideal” customer should be. They rarely get tested against what actually happened in the pipeline. This framework builds an ICP the other way round: starting from deal history and observed buying signals, then tiering accounts by how well they match. It’s the same sequence we use in every ICP consulting engagement and GTM Diagnostic.

Why deal-history-first beats assumption-first

A firmographic ICP (“50–500 employees, US or UK, Series B+”) describes a company shape. It says nothing about whether that company has a reason to buy right now. Two companies with identical firmographics can have completely different buying readiness — one just changed leadership and is actively evaluating vendors, the other renewed a competitor’s contract eighteen months ago. Deal-history-first ICPs surface that difference; firmographic-only ICPs can’t.

Step 1. Pull deal history

Gather 12 months of data (or as much as exists) across four buckets: closed-won, closed-lost, churned and renewed accounts. Include deals that never should have entered the pipeline in the first place — those are often the most instructive, because they show where targeting drifted from what actually converts.

If you have fewer than roughly 20–30 closed deals total, treat this framework as directional rather than statistically reliable, and weight qualitative sales/CS interviews more heavily in Step 2.

Step 2. Identify differentiating signals

Beyond size, industry and region, the accounts that convert fastest and retain longest usually share signals that a firmographic filter won’t catch:

  • Operational triggers — a leadership change, a funding event, headcount growth in a specific function, or a recent tool rollout that created a new gap
  • Budget reality — evidence of existing spend allocated to the category, not just a plausible use case
  • Readiness to adopt — technical maturity, internal champions, or a documented mandate to solve the problem this quarter rather than “someday”

Pull these from CRM notes, deal-loss reasons, and 15-minute interviews with the reps who ran the deals — the signal is often in a rep’s memory of “why this one moved fast” long before it’s in any dashboard.

Step 3. Build a tiered model

Score every closed and open account against the signals from Step 2, then group into three tiers:

  • Tier 1 — strongest combination of firmographic fit and readiness signals. These accounts should get the most outbound attention and the fastest sales follow-up.
  • Tier 2 — firmographic fit is present but readiness is unclear or unconfirmed. Worth targeting, but expect a longer cycle and lower conversion than Tier 1.
  • Tier 3 — firmographic fit only, no observed readiness signal. Deprioritize for cold outbound; these accounts convert better through inbound or partner channels where they self-select.

Step 4. Test against current pipeline

Score every open opportunity against the same tiers. This is the step most teams skip, and it’s the one that actually validates (or invalidates) the model. A high concentration of open pipeline sitting in Tier 3 is a concrete, measurable sign that outbound targeting has drifted from the ICP — not a hunch, a number you can put in a slide.

Step 5. Revisit every 2–3 quarters

An ICP isn’t a one-time deliverable. Revisit it after a product launch, a move upmarket or downmarket, a new competitive entrant, or a noticeable shift in win rate. Teams that treat the ICP as a living model — reviewed on a cadence rather than left in a slide deck — catch targeting drift months before it shows up in missed pipeline targets.

Common mistakes that undo this framework

  • Scoring on firmographics alone and calling the result an ICP. Firmographics describe fit; they don’t describe readiness. Both are needed.
  • Never revisiting Tier 3. Sometimes Tier 3 volume is fine — it’s a deliberate top-of-funnel or partner-led segment. The mistake is not knowing which reason applies to your pipeline.
  • Building the model from marketing personas instead of deal data. Personas describe who you imagine buying; deal history describes who actually did.
$300 · 60 minutes

Start with a GTM Diagnostic.

60 minutes. One clear diagnosis. A practical next-step plan.