Item-Level Reviews and Ratings to Improve Zomato Menu Conversions
How item-level dish reviews could take the guesswork out of ordering from an unfamiliar restaurant on Zomato
✳The problem they went after
Working professionals ordering on Zomato can see a restaurant rating but nothing about the individual dishes. Ordering several items from a place they do not know is guessing on taste and portion size. That uncertainty stretches out menu browsing and shows up as drop-off between menu views and cart additions.
✳What they actually did
The proposal is a post-delivery review prompt. Forty-five minutes after the order lands, the app asks the user to rate each dish one to five stars and optionally tag it 'too spicy', 'great taste', or 'generous portion'. Those verified ratings get aggregated onto the menu item itself. It beat restaurant-supplied nutritional data and third-party review links because Zomato ends up owning the trustworthy database. Review completion rate is the primary metric, targeting 15 percent in 60 days, with menu-to-cart conversion up 5 percent in 90 days as the business outcome.
✳What you can take from it
Worth stealing
- Post-delivery prompt timed 45 minutes after the order arrives
- Structured tags like 'too spicy' instead of free-text reviews
- Three-review threshold before an item rating goes public
- Fraud detection named as a guardrail against restaurant-planted reviews
Read this if
Anyone designing a UGC loop that needs volume before it becomes useful
The transferable bitWorth studying for how the metric stack is built. There is a behaviour metric the feature directly controls, a business metric it should move, and guardrails that keep both honest: fraud detection against restaurant-planted reviews, and no public rating until an item has at least three. The guardrails are where the real thinking shows.
✳The submission itself
Open the PDF →Reading on a phone? The embedded viewer is cramped — open the PDF in a new tab instead.
