ICP and Buyer Modeling

Owned boundary and outcome

This lesson produces ICP Buyer Evidence Matrix v1. It does not build a flattering persona. It bounds each segment by market evidence, buyer role, problem intensity, decision authority, budget path and real constraints. Final serviceability and lead scoring remain adjacent outcomes.

Problem to solve

Teams often treat demo enthusiasm, demographic resemblance or a few friendly interviews as proof of fit. That shortcut hides who experiences the problem, who pays, who can decide, what triggers purchase and which procurement constraint can stop the deal.

FACT, INFERENCE and DESIGN

SBA market-research guidance and the Salesforce qualification example are external evidence boundaries only. They do not define a universal threshold or MAKNOW state. Separating market evidence from buyer-path evidence is an instructional inference; MAKNOW states are local design.

ICP evidence contract

For each segment, evidence should expose buyer role, problem intensity, decision authority, budget path, procurement constraint, current alternative, evidence type, confidence, exclusion reason, review trigger and evidence reference. UNKNOWN must stay visible rather than being filled with guesses.

Decision states

SUPPORTED means the current hypothesis has adequate evidence. HYPOTHESIS means decisive evidence remains weak or incomplete. EXCLUDED means a valid current constraint removes the segment. HOLD means evidence is insufficient for a decision. PARTIAL, DISPUTED, STALE and BLOCKED preserve other bounded non-success conditions.

Primary matrix

The table binds the decision to auditable evidence.

Defensible baseline

A segment with recurring economic pain, a real decision maker, a clear budget path, no blocking procurement constraint and multi-source evidence is SUPPORTED. Interest or contact count alone is not enough. Each row should show why the problem recurs, how buyer and user differ, and how the budget path reaches the decision.

Failure mode: champion without authority

A champion may explain the pain well but still lack purchase authority or a viable budget path. The state should not exceed HYPOTHESIS unless other evidence identifies the real decision path.

Failure mode: sample bias

Interviews drawn from friendly customers, current users or one channel can create sample bias. Bias does not automatically exclude the segment, but it reduces confidence and requires an explicit review trigger and additional evidence.

Constraints and exclusion

Procurement, security, regulation or contract can make an attractive segment currently EXCLUDED. Exclusion needs a reason, evidence reference and review trigger. Planned future relief is not current evidence.

Execution workflow

1. Name the segment. 2) Separate market evidence from buyer evidence. 3) Identify user, champion and decision maker. 4) Record problem, trigger and economic impact. 5) Check budget, procurement and alternatives. 6) Record confidence and bias. 7) Assign state and review trigger.

Offline lab

The lab contains eight synthetic fixtures. Each fixture checks eight deterministic outputs: state, reason, authority, budget path, constraint, evidence quality, missing-critical count and recompute-required. No live customer data or API is used.

Transfer

After a regulation removes a procurement constraint, that segment can become SUPPORTED; a former champion who loses authority must be recomputed and may fall back to HYPOTHESIS. The baseline is not an oracle; changed regulation or authority requires independent recomputation from changed evidence.

Assessment

You must explain the ICP boundary, build the matrix for the baseline, diagnose a champion without authority, and recompute states in the transfer case. A positive label without traceable evidence is insufficient.

Review and stop/pivot triggers

Each segment needs a review trigger: regulatory change, buyer-authority change, budget-path change, new evidence of bias or stale evidence. New hard constraints or conflicting evidence must weaken the state instead of preserving the old persona.