Weekly Active Users
142.8K
↑ 12.4% WoW
Match → Click Rate
38.2%
↑ 4.1pp WoW
Click → Apply Rate
24.6%
↑ 2.8pp WoW
Apply → Interview
8.4%
↑ 0.9pp WoW
Discovery → Match → Apply → Interview Funnel
7-day window · all users · synthetic
Discovery
100.0%
—
AI Match Surfaced
84.1%
−15.9%
Match Clicked
38.2%
−45.9%
Application Started
42.0%
−58.0%
Application Submitted
58.7%
−41.3%
Interview Confirmed
8.4%
−91.6%
Largest Drop-Off
Discovery → Match Click loses 74,257 users (−51.9%). The biggest lever is match score visibility on the discovery feed. Surfacing top-3 score reasoning in the card preview drove +8pp click rate in the v3 cohort.
Conversion by Match Score Band
Apply rate by AI confidence score
Match Score Distribution
All matches surfaced this week
Agent Recommendation Accuracy
82.4%
↑ 3.1pp vs prior model
Avg Match Score
78.2
↑ 4.8 WoW
Median Latency
412ms
↓ 89ms WoW
Flagged for Review
128
12 require retraining
Agent Performance by Match Type
Quality scoring · auto + user feedback
| Agent | Volume | Acceptance | User Rating | Latency p95 | Status |
|---|---|---|---|---|---|
| SkillMatch v3.2 | 48,290 | 87.4% | 4.6 / 5 | 684ms | PRODUCTION |
| RoleFit v2.8 | 31,124 | 81.2% | 4.3 / 5 | 512ms | PRODUCTION |
| CompanyCulture v1.4 | 22,841 | 78.9% | 4.1 / 5 | 892ms | REVIEW |
| SalaryRange v2.1 | 18,432 | 74.6% | 3.9 / 5 | 324ms | PRODUCTION |
| VisaFilter v1.0 | 14,289 | 68.4% | 4.4 / 5 | 218ms | PRODUCTION |
| RemoteAware v0.9 (beta) | 8,124 | 62.1% | 3.6 / 5 | 1.24s | RETRAIN |
Acceptance Rate by Agent
Last 30 days
Latency Trend p95 · 14d
Time to first match · ms
Active Experiments
7
3 reaching significance
Apply Rate Lift (Best)
+18.2%
Variant B · p=0.003
Avg Sample Size
14.2K
Per variant
Tests Shipped (Q1)
23
8 winners promoted
Experiment: Match Card Reasoning Display
Hypothesis: surfacing top-3 match score reasons increases apply rate
CONTROL · Variant A
22.6%
Match score shown only · no reasoning surfaced on card
TREATMENT · Variant B WINNER
26.7%
Match score + top 3 score-driver bullets shown inline
Lift
+18.2%
P-value
0.003
95% CI
[+8.4%, +28.0%]
Power
0.94
Recommendation
Promote Variant B to 100% of traffic. Significant lift, sufficient power, no negative impact on downstream conversion or retention. Estimated annualized impact: +1.4M additional applications at current weekly active user volume.
Currently Running Experiments
Status as of Apr 27, 2026
| Experiment | Variants | Sample | Lift | P-value | Status |
|---|---|---|---|---|---|
| Match Card Reasoning | A vs B | 28,430 | +18.2% | 0.003 | SHIP |
| Resume Tailoring CTA | A vs B vs C | 42,891 | +9.4% | 0.018 | PROMOTE |
| Onboarding Flow Length | 3-step vs 5-step | 18,242 | +6.1% | 0.082 | RUNNING |
| Visa-aware Filter Default | On vs Off | 12,489 | +4.2% | 0.124 | RUNNING |
| Push Notif Timing | 9am vs 6pm | 28,140 | +1.1% | 0.412 | NULL |