Cocoon
Intelligence / platform

Flags & experiments

28 flags · 6 experiments running · 2 ready to ship · 1 needs a decision
AllRunningReadyArchived

Experiments

Primary metric · significance
ExperimentTrafficPrimary metricLiftpState
match-v4.3 weights12%Match → interview+2.5pt0.01Ready
Dot prompt v1912%Call completion+1.8pt0.04Ready
Voice-first role posting50%Roles posted / week+22%0.00Running
Shortlist of 5 vs 1050%Time to shortlist−0.6d0.11Running
Skills-first onboarding30%Profile completion+6pt0.02Running
Salary transparency nudge20%Application rate+9%0.06Decision
Min sample 4,000 per armTwo-sided, α 0.05

Flags

28 total
FlagRolloutOwnerState
dot_voice_v3100%E. RoweOn
gem_surfacing100%M. PatelOn
portal_dark_theme35%R. SmithRamping
ats_bulk_export9 acctsD. KimBeta
auto_verify_pass61%T&SRamping
offer_esign0%D. KimOff
candidate_referrals5%M. PatelBeta
Owner required per flagStale > 90d: 3

Decision needed

Salary transparency nudge
Application rate up 9% but company posting rate down 4% — trade-off needs a call from product and trust.
Applications
+9%
Roles posted
−4%
Candidate trust score
+0.3

Release log

Last 6
match-v4.2 to 100%14 Jul · M. PatelDiff
Dot prompt v18 live11 Jul · E. RoweDiff
Verification auto-pass to 61%8 Jul · T&SDiff
Portal voice posting to 50%4 Jul · R. SmithDiff
Rollback: offer_esign2 Jul · DocuSign 500sDiff

Guardrail metrics

Auto-halt if breached
p95 latency < 800ms
620ms
Fairness parity > 0.85
0.93
Call completion > 90%
94%
Support volume < 20/day
12
Error rate < 0.5%
0.11%