Yelp (Demo Mode) · Consumer Marketplace
New restaurants joining Yelp are producing fewer reviews in their first 14 days, threatening local marketplace liquidity.
Executive Summary
Talk to usCost of inaction · 30 days
~0fewer reviews
2 of 4 active segments are on a high-risk trajectory over the next 14 days — if nothing changes.
Largest projected loss · Early Review Drop-off
First Review Rate · 41% → 36.5%
Focus firstEarly Review Drop-off (largest projected loss).
of 6 discovered · 2 below signal threshold
Risk 70+ and decline probability 60%+
First projected escalation
Yelp (Demo Mode)
Top Projected Risk Segments
Ranked by risk score and decline probability. Click a segment for the full analysis and the audit trail.
Risk by horizon
Risk score per cohort at each forecast horizon, worst 30-day first. 80+ critical · 60–79 high · 35–59 moderate · under 35 low.
| Cohort | 7d | 14d | 30d |
|---|---|---|---|
90 | 97 | 100 | |
61 | 66 | 72 | |
14 | 11 | 8 |
Early Review Drop-off
CRITICALRISK 97Consumers on iOS in NYC are leaving fewer first reviews on newly listed restaurants. The simulation projects the first-review rate to keep declining over the next 14 days, driven by verification gating on new listings and added steps in the iOS review-submit flow.
Cost of inaction · 30d
~18,770 fewer reviews
Android Re-engagement Decay
HIGH RISKRISK 66First-review rate for new LA listings is softening on Android, where reduced notification delivery on Android 14+ is weakening the consumer re-engagement loop that historically prompted early reviews.
Cost of inaction · 30d
~1,407 fewer reviews
Supply-Side Resilience
RISK 11Review receipt on established listings in top metros is stable to slightly improving. This cohort acts as a stabilizing force on marketplace liquidity and shows no near-term decline risk.
Cost of inaction · 30d
~0 fewer reviews
Bands failed backtesting — ranked by projected loss
For these cohorts, reality landed inside our predicted range less reliably than we promise. The trend and the projected loss are computed differently and still stand, but we withhold the risk score and every probability rather than dress up a range we could not defend.
Web Activation Stall
HIGH RISKRISK WITHHELDThe direction below still holds — what we could not defend is how confident to be about it. Reality landed inside this cohort's predicted range 97% of the time in backtesting against a target of 80%.
Web consumers in NYC convert to a first review at a lower rate than mobile, and the trajectory is drifting down. No mobile-web re-engagement path means consumers who view but don't review a new listing rarely return.
Cost of inaction · 30d
~1,676 fewer reviews