As Zomato became the go-to app for foodies around the world last year, we began to see a rise in biased content on our platform – most of this appeared to come in from various social media agencies. Meanwhile, a lot of users had also taken to writing reviews that weren’t particularly constructive, in a bid to climb the Leaderboards in their respective cities.  

All ratings on our site are governed by user activity, so ensuring that these are generated by fair and neutral user inputs is of paramount importance to us.

Our first big step towards countering these reviews and ratings was Project Warp – a complete overhaul of our ratings algorithm along with a few other essential bias checks being put into place. Here’s the long and short of what we did and how it helped –

• User credibility scores were updated. This helped us more accurately predict individual “bias” in reviews.
• The recency of reviews matters a whole lot more now, and prevents restaurant ratings from being drastically affected by very old reviews.
• Reviews that have been moderated for policy violations such as abuse or plagiarism automatically carry less weight.

We have a detailed blog post, with a full breakdown of how Project Warp works. You can read it here.

Too much of a good thing

We originally introduced the concept of Leaderboards to honour our users who shared reviews and photos regularly on Zomato. Some of our users were even writing almost 30 reviews a day! We were super excited about this flurry of activity, but we later found that this healthy “gamification” had resulted in reviews being plagiarised. Not to mention a lot of “filler” reviews flooding the platform, that weren’t particularly helpful to other users. Even more disconcerting, was the continued trend of users leveraging their position on the Leaderboard to solicit restaurants, something we’ve been fighting since day one. But what hurt us the most was our users telling us that Zomato reviews weren’t particularly helpful when looking for a restaurant.

Reviewing our review system

We’re now taking a few steps to combat this. Starting today, our users in India will be able to write a maximum of 10 reviews per month. We know, you’re probably thinking, “These guys are utterly insane!” but bear with us, and read on:

• The aim of this new review limit is to help users build credibility on the platform through better content over a period of time, rather than them being able to become influencers practically overnight.
• We will also be phasing out review and blog Leaderboards. Given that frequency of activity was the key factor behind earning a spot on the Leaderboard, this no longer seems to be appropriate recognition, with the new 10 review limit. Inputs from users and restaurants, who felt the Leaderboard was enabling a lot of spam on Zomato, helped solidify this decision.
• That being said, we are already in the process of re-imagining and reworking Leaderboards, so that users can be appropriately rewarded for both quantity and quality.

What this means to you as a user

The call to introduce this new review limit was a data-informed one. We crunched the numbers thoroughly in order to measure the kind of impact the decision would have on our sizeable India user base.

• Almost 97% of our users in India will be able to carry on writing reviews as they did before, and won’t be affected by the change at all
• We do understand that the limit may be a problem for power-users who have been active on Zomato for a few years now. We’d like to ask you to please write to us at neutrality@zomato.com, if you have any concerns about this change

We completely get that this change is hard to take, but ensuring that site remains relevant across the board for our user base, is vital. Zomato’s core mission is helping people identify great places to eat. Reviews form a significant chunk of the decision-making process, and we need to keep doing what we can to safeguard the relevance and usefulness of our content for every foodie out there.