Case Study: Whatnot scales from 0 to a ~400-per-year experiment run rate with Statsig
Key results
The challenge
Whatnot, a livestream shopping marketplace where over 200 products sell every minute, wanted to build a rigorous experimentation practice. It needed results the whole organization could trust and a platform that could scale with its data.
The solution
Whatnot rolled out Statsig, moving to Warehouse Native experimentation and tuning technical parameters such as its CUPED lookback window from 7 to 14 days for meaningful variance reduction on key metrics. Feature flags and canary releases supported safer launches.
The results, in context
Whatnot went from 0 to 250 experiments over three quarters, a run rate of roughly 400 experiments annually. The company reported gains in gross merchandise value and activation rates as experimentation scaled, against a backdrop of more than 200 products sold per minute.