Separate stage from score

A lifecycle stage describes a business state; a score estimates propensity or priority. Combining them creates records that jump stages when a score changes and makes funnel history unreliable.

Use a short set of states that sales, marketing, success, and finance can define the same way.

Make transitions event-based

Record the current stage, the transition timestamp, prior stage, actor or automation, and reason. Avoid reconstructing history from a mutable current-value field.

Define whether stages may move backward, when disqualification is terminal, and how reactivation starts a new cycle.

  • Entry condition
  • Allowed prior states
  • Timestamp
  • Owner
  • Reason
  • Reversal policy

Audit the funnel denominator

A conversion rate depends on the cohort entry rule and observation window. Report open outcomes separately from failures rather than treating unfinished records as lost.

Monitor stage age and impossible transitions as data-quality signals.

Verification checkpoint

Choose a sample of records from each stage and prove the current value and transition history match the written entry rule.