Aman Priyadarshi | September 20, 2019 | 4 min read
Reviews 2.0 – tags are a new content currency

In our last annual report, we talked about how deeply we care about “better food for more people” and how it is the very DNA for everything we do here at Zomato.

That’s exactly where Zomato started out — a platform to discover restaurants which had good reviews so that you can have a great experience. Whether food quality, ambience, cost, service or even the crowd that walk into their outlets — restaurant owners received 360 degree feedback from users. Outlets which worked to improve their critical reviews saw their rating go up and in return drew more business. On the other hand, customers were happy being able to choose better places (with higher ratings) to go to. Win-win.

A decade later — we have over 25m reviews and 52m photos for about a million restaurants listed on Zomato across the world. This makes us one of the largest food discovery platforms in the world. But here’s the catch — all these reviews were written by only 1% of our monthly active users (MAUs) on Zomato. Which means that for every 100 users who use Zomato to consume reviews, only 1 user writes a review.

Over the next 10 years, we want a lot more users to write reviews and share their perspective with the large community that we have built.

Reinventing Reviews, but where do we start?

If there’s a demand, supply follows. So, we went back to our users who solved it for us.

We asked people why didn’t they read as many reviews and here’s what we found –

Clearly, our current supply of reviews didn’t adhere to the evolved demand. To make people read reviews, we had to make reviews crisper than they currently were.

Now that we had a better understanding of what kind of reviews people want to read, we still didn’t know what was stopping people from writing more reviews. Hence, we asked our users what kept them from writing reviews frequently?

With this, our takeaway was ready.

  • Present reviews in a short and crisp manner so that they are easy to consume
  • Help people to write great reviews with less effort

How did we design our solutions?

Our designers are some of the best in town. Given a problem, they come up with great answers in high fidelity.

This is what they conjured up after multiple iterations –

The 5 tags under “what do people say about this place” have been parsed out of 1091 reviews that we have received for this place. Would you believe that you read over 1091 reviews for an outlet in the image above? You really did!

Our user testing showed that users loved this format of summarising the content of reviews in form of tags for them; so we thought that we should introduce this new content currency of tags to the creation of reviews.

In Reviews 2.0, you don’t have to be a pro reviewer to write good reviews. You can now add a very helpful review on Zomato in under 30 seconds.

You can choose to not write a long-form text review. You can simply tap on a suggested tag or add a new one. Thus a good quality yet crisp review can be written in a breeze!

Now that we had solved for both creation and consumption of reviews at a design level, it was time for our engineering team to take it to the finish line.


How did we develop Reviews 2.0?

Truth be told, the perfect user experience had compelling engineering and data challenges before Reviews 2.0 could see the light of the day. Here were the top problems we had to solve for –

Challenges for ‘Create Review’ –

  • We would need a library of tags to save users time while writing reviews (tapping is faster than typing + it wouldn’t require a second moderation so does away with spelling mistakes)
  • In order to support good quality review creation, the suggested tags would have to be relevant to that outlet

Challenges for ‘Read Review’ –

  • How would we show tags on the restaurant page on day 0?
  • How would we make review tags backward compatible with our existing reviews?

We worked on multiple strategies, using solutions across data aggregation and machine learning models to solve for the above challenges — you can read all about it in our next post on Reviews 2.0 (coming up soon).


How are Reviews 2.0 performing?

At 30% app adoption, Reviews 2.0 already contribute 60% of our daily reviews and are projected to increase the rate of review creation by 100% overall (win!).

Moving forward, Reviews 2.0 opens a lot of avenues for our platform such as –

  • Better search – Search anything on Zomato, not just cuisine, dish or restaurant name. You will soon be able to search experiences – ‘perfect for conversation’ or ‘great for family dinner’
  • Enable match % – Predict if a user will like or not like a place even if the user has never been there

Have you tried writing a Review 2.0 on Zomato yet? If not, do download the latest version of the app from the app store and give it a spin.

facebooklinkedintwitter

More for you to read

Technology

apache-flink-journey-zomato-from-inception-to-innovation
Data Platform Team | November 18, 2024 | 10 min read
Apache Flink Journey @Zomato: From Inception to Innovation

How we built a self-serve stream processing platform to empower real-time analytics

Technology

introducing-pos-developer-platform-simplifying-integration-with-easy-to-use-tools
Sumit Taneja | September 10, 2024 | 2 min read
Introducing POS Developer Platform: Simplifying integration with easy-to-use tools

Read more about how Zomato is enabling restaurants to deliver best-in-class customer experience by working with POS partners

Technology

migrating-to-victoriametrics-a-complete-overhaul-for-enhanced-observability
SRE Team | August 12, 2024 | 11 min read
Migrating to VictoriaMetrics: A Complete Overhaul for Enhanced Observability

Discover how we migrated our observability metrics platform from Thanos and Prometheus to VictoriaMetrics for cost reduction, enhanced reliability and scalability.

Technology

go-beyond-building-performant-and-reliable-golang-applications
Sakib Malik | July 25, 2024 | 6 min read
Go Beyond: Building Performant and Reliable Golang Applications

Read more about how we used GOMEMLIMIT in 250+ microservices to tackle OOM issues and high CPU usage in Go applications, significantly enhancing performance and reliability.