Rating bias: Zomato CEO Deepinder Goyal proposes ‘match score’ over crowd opinion, sparks debate

Zomato CEO Deepinder Goyal has proposed a shift in how users discover restaurants on the platform. Instead of relying on traditional ratings, Zomato is experimenting with a personalized "match score" — a system that recommends eateries based on individual taste preferences rather than mass reviews.  

In a post on X (formerly Twitter), Goyal explained the reasoning behind the move. “We all have different tastes in food — so why rely on the same restaurant ratings? Internally, at Zomato, we have been trying out personalised ‘match scores’ instead of traditional restaurant ratings, and we’re loving it.”  

He further elaborated, “Match score reduces bias from mass opinions that might not match your taste. As a result, we’re discovering more relevant restaurants than ever before. But we’d love to hear from you! Would you prefer match scores tailored to your preferences or stick with traditional ratings?” 

The proposal sparked mixed reactions. Some users welcomed the idea, saying it could lead to better dining experiences. “That sounds like a brilliant idea. I’m always frustrated when a highly-rated restaurant just doesn’t align with my personal tastes. Personalized match scores could definitely help me discover hidden gems that I might otherwise miss. I'm all for it!” wrote one user.  

Others, however, raised concerns about the over-reliance on algorithms and the potential for commercial bias.

“The problem with match score is over-reliance on algorithms and control of the corporation. You might not do it today, but tomorrow you can take advertising money and match more. Good for marketing but bad for new entrants. Have both at disposal to the customer,” commented another.  

Some feared the approach could lead to "familiarity bias", limiting exposure to new experiences. “You shouldn’t do match scoring algorithms - this is the classic case of splinternet - a fragmented network that is only available to me based on who I am and my behaviour. It will create familiarity bias.”  

Others advocated for a hybrid model that combines traditional ratings with personalized recommendations. “Can be both and love the idea of personalized match score! Can show ratings, personal match (similar to Netflix), and what’s trending as well (top 10 in Chandni Chowk, for example),” suggested a user.  

Another echoed this sentiment, saying, “The idea is great because food is deeply personal. IMO, a match score could help uncover hidden gems that align with individual tastes rather than just popular opinion. That being said, I wouldn't rely entirely on a match score alone. A balanced approach would be blending both traditional ratings and personalized match scores.”