Winner Detection

Composite score per (advertiser, product). Replicates your manual workflow: re-investment burst + lifecycle + storefront rank + cross-advertiser.
How the score works (click to expand)
Composite weights: horizontal_scale=1.0 · active_ratio=2.0 · recent_burst=3.0 ★ · lifecycle=2.0 · country_match=1.5 · cross_advertiser=2.0 · multi_country=−1.0 · store_rank_top3=4.0 ★ · store_rank_top10=2.0 · reach=0.5

Tier bands: S score ≥ 12 AND ≥2 ads in last 7d   A score ≥ 7   B score ≥ 4

Re-investment burst = # ads with start_date in last 7 days. The KEY signal — when an advertiser came back and uploaded more ads, that's the "they're scaling it" signal you described.
Storefront rank = the product's position when sorting /collections/all?sort_by=best-selling on the advertiser's own store. Top-3 = they're literally selling it the most.
Total ranked
S-tier
A-tier
With store-rank
Today's burst (∑)
Loading winner scores…