Real-time recommender for Amazon sales

Problem

Sales teams couldn’t spot anomalies and opportunities buried in Amazon sales data — lost sales, loss-on-sale, price-matching issues, competitor stock-outs, traffic spikes. The signal was there; nobody had time to find it.

Approach

I built a real-time ML alert/recommender that detects these signals as they happen and recommends the best action to take, with an additional model to catch overlooked competitor cannibalization.

Result

Activated across 5 brands (Braun, Gillette, Venus) in 6 countries, feeding day-to-day decisions for sales executives.

Notes

I was Product Owner for the tool, led a small team, and balanced conflicting requirements across departments.

Screenshots coming soon.