Why do hidden ICP segments form inside a customer base?

The official ICP is a filter built from early customers, who are themselves a biased sample of the broader market. Some non-ICP customers slip through and close, but they are dismissed as exceptions and never aggregated as a segment. Over time, a small but real cluster accumulates, hidden because the analytical frame is bound to the existing ICP.

What is anomaly mining and why does it find hidden segments faster?

Anomaly mining looks at deals that should have lost but won, customers who should have churned but stayed, and accounts that expanded beyond predictions. Each surprise points at a hidden factor not captured in the official ICP. Mining 20 anomalies surfaces candidate segments much faster than full corpus analysis.

How do you operationalize a validated hidden segment?

Treat it as its own motion: dedicated target list, defined messaging, sequence design specific to the segment, and reporting tagged to it. Resourcing depends on segment size, but the wrong pattern is to dilute it into the main motion, where segment-specific work loses to the main ICP motion by default.

Finding Hidden ICP Segments Through Research

Most teams have ICP segments hiding inside their customer base that the official ICP never described. Structured research surfaces them and opens new growth without product changes.

Why Hidden Segments Exist

The official ICP is a filter that excludes data before analysis. Non-ICP wins accumulate as exceptions and stay invisible because the reporting frame is bound to the existing ICP.

Research Method

Corpus assembly, unsupervised clustering, hypothesis articulation, and sizing produce candidate hidden segments that can graduate from interesting clusters into addressable opportunities.

Anomaly Mining

Won-but-shouldn-have-lost deals and stayed-but-should-have-churned customers point directly at hidden factors. Mining 20 anomalies surfaces candidate segments faster than full corpus analysis.

Validating and Expanding

Build a signal pattern, run a focused pilot, validate persistence with a second pilot, then expand into the segment as its own dedicated motion with separate reporting.