ICP and segmentation

Use this when the user asks who to sell to, names a segment, or hands over customer lists, win-loss notes, or pipeline data for a targeting read. Start from the customer or pipeline data on hand — HubSpot, a sheet, or a pasted export — the ICP hypothesis in memory, and the decision the read informs.

Define the ICP in six blocks

Firmographics with technographics (size, industry, geography, the stack they run), the trigger event (the situation that makes them buy now), personas (buyer, champion, blocker), capabilities they must already have, fit signals, and anti-signals that disqualify. Build from best customers — retention, deal size, health, advocacy — never the whole base; largest is not best. Quote the evidence behind each block or mark it UNKNOWN.

Fit is not ready

Split fit accounts from ready accounts: an intent layer (behavioral and surge signals) plus budget and use-case fit sits on top of firmographics, so targeting chases companies ready to buy, not just shaped to buy.

Tier the base

Strategic, growth, and watch. Each tier gets why it is there, the motion it earns, and the graduation or kill bar that moves accounts in or out.

Size each segment

Count, coverage, win rate, and deal velocity where the numbers exist. When Apollo is connected, count the accounts that match the firmographic block so the tier has a real denominator. A segment you cannot count stays a hypothesis, said plainly.

Name what you are not chasing

The deprioritized segments, out loud, with the reason — weak fit, no access, unwinnable against the incumbent — so sales stops spending cycles there.

What you hand back

The ICP memo with segments and targeting priorities: the ranked account or lead list shape, the qualification bars per tier as a Best/Good/Bad Fit rubric for routing, and the handoff note for sales and marketing to execute against. You design the target; they run the motion. Save the memo dated with sources. Revisit it quarterly — an ICP goes stale after six months — and sooner when win-loss themes shift or a tier misses its bar two periods running.

Fallbacks

No data means you draft the ICP from their best-customer stories and mark every line INFERENCE. A thin list means tiers with graduation bars, not false precision. Never invent a customer, a segment size, or a win rate, and never brief the same list twice — fold in what the scorecard routine already logged.