Use this when a team needs to measure whether CRM reliability is improving. Produce a reproducible score and diagnosis, not a decorative grade.

Define health as rules

Translate “clean CRM” into explicit checks across identity, ownership, completeness, pipeline integrity, process conformance, and integration drift. Every rule must declare its population, condition, severity, evidence, exclusions, and owner.

Measure only fields and relationships that influence execution or decisions. Do not reward filling optional fields merely because they are easy to count. Keep rule definitions versioned; a changed rule creates a new baseline and must not be presented as organic improvement or decline.

Score the population

Calculate rule failure rates on eligible records. Weight by business impact, not raw finding count. A handful of duplicate accounts splitting live opportunities can matter more than thousands of missing optional fields.

Score objects separately before rolling them up. Contacts, accounts, opportunities, and activities have different populations and failure modes. Use references/scoring-model.md for a deterministic 0–100 model and minimum sample rules.

Report the score with:

crm-health-scoring

Define health as rules

Score the population