WorkMango Match
Mango Match
In 2023 the apps team tried something ambitious: using play to train the recommendation algorithm. Turn discovering clothes into a quick, fun game, and get something we never had before in return, explicit feedback on what people liked, and what they didn’t.
- +28%Average session time vs. customers who didn’t play
- +7%Add to bag from the final recommendations list
- +35%Recommendation accuracy for customers who played
This case study includes confidential details
The summary and results above are public. The rest, how the project actually worked, is shared under a password because it covers information Mango hasn’t made public. If you’re a recruiter or hiring manager and don’t have it, get in touch and I’ll send it over.