Freeze the population before seeing the hard cases
Internal commercialization control. A 3–5-event serviceability cohort is invalid if clean or inexpensive events are chosen after detailed evidence review.
Selection
Freeze the population of eligible completed events for one provider/process/form family using metadata only. With 3–5 events, use the census. With more than five, use the most recent five completed eligible events.
Do not replace selected events
Missing evidence, blocked access, factual exceptions and unexpectedly high review effort remain part of the measured outcome. They are not reasons to swap in an easier event.
Boundary
This slice is not statistically representative and does not estimate population exception rates or future performance. It creates no audit, assurance, value, WTP, price or financing conclusion.