# Cohort and experiment instructions

Count unique contractor-engagement onboarding episodes, with repeated invitations deduplicated. Use the first valid invitation as the entry time. All example participants have 30 complete elapsed days of observation. Count a milestone only if it occurs within those 30 days and follows the defined sequence. Keep returning-contractor episodes separate if their workflow differs.

Stage conversion equals reached count divided by previous-stage count times 100. Overall activation is 72 / 200 = 36%. Stage counts are cumulative and overlap; never sum 200 + 170 + 150 + 130 + 90 + 72 as people. First-stage conversion has no previous-stage denominator and is blank.

The 40 ready-but-unassigned people equal 130 - 90. The 20 submitted-but-not-ready people equal 150 - 130; synthetic reasons divide into 10 internal-review cases and 10 eligibility decisions. Don't combine unknown, ineligible, no-offer, and contractor-withdrew reasons into abandonment.

The experiment row is a planning example, not an executed test or claimed result. Define assignment and guardrails before starting. Compare mature cohorts with consistent definitions and report incomplete cohorts separately.