How Not To Build Your Rating and Reviews System

When enabling a transaction becomes commodity, the value of marketplace lies in enabling trust. Trust that the marketplace will connect the consumer with worthy service provider. Feedback from past users are the key to building this trust. I have written about this in detail back in 2013 for various industries.

There are few practices in feedback collection that are being overlooked by popular marketplaces that will end up creating less trustable profiles in the long run.

  • Forced and Untimely feedback: From what I recall, Uber started this practice of collecting feedback about previous transaction at the beginning of new transaction. While it is designed to collect feedback at the end of the current transaction, users have no reason to look at the app in the end. What are the chances that I actually remember my last transaction unless it was horribly bad? What are the chances that I want to pause for a while and put efforts to give a genuine rating. At this moment, the feedback screen is just a pop up with a skip button camouflaged in the rating stars. I wouldn’t want to sound too judgemental and there is always room for improvement so I give a 4 star. Most users end up giving 4 or 5, because thinking, at that moment when you are waiting to book a ride, is too much cognitive load.
    While cab rides are commodity and all factors are hygiene factors (from Herzberg’s theory, not from the Dettol ad), this practice didn’t quite seem right when experiencing on Zomato’s Delivery orders.Solution: Have a time limit. If the feedback isn’t submitted by then, don’t ask. The time limit should depend on the impact of service. A holiday experience could have a month’s time limit but for a cab ride, probably not more than 24hrs.
  • Value of Individual feedback: If a person gives only between 3-5 stars everytime would you treat his 3 same as that of a person who only gives between 1-3 stars. If a feedback was given 1 year ago would it still be treated same as the one that was given yesterday? Zomato did a great job with this, described in detail on their blog.
    Lately, Zomato has started treating delivery orders’ rating at par with dine-in’s rating. The basic problem with this treatment is that these 2 are very different services with food being the only common aspect out of 5. For dine-in, I would value service and ambience as well. Whereas for delivery, the packaging and delivery time would also be valued. Now by looking at a rating on Zomato it is difficult to say what this restaurant is good at. By design and the point mentioned earlier, a restaurant that does more delivery will continue to get better ratings.
  • Representation of feedback: How reviews and ratings are shown helps make best use of it. Showing who gave the rating (couple on honeymoon or solo traveller), when was it given, what was the rating specifically meant for (product or delivery) etc. helps comprehend better. It is also important to differentiate between what part of the feedback is for the service provider and what is for fellow users.

Should a transaction platform care about ratings and reviews? Ratings are an important aspect for curation. What comes on top of search and collection pages is determined by rating. A misinformed rating creates wrong expectation and that leads to a bad user experience. Remember, payments and delivery are a commodity, trust is the only differentiator.


  1. Tripadvisor does a great job with reviews. Pretty detailed. Focus is on quality and not quantity.
  2. Youtube used to have a 5 star ratings system. Now a binary system. Lesser cognitive load.
  3. Flipkart web app used to ask ratings on past orders in a widget on the side bar, never forced over a pop-up.
  4. Avoid using reviews system as a way to build content for SEO. A QnA section can do that job better.

Leave a Reply

Your email address will not be published. Required fields are marked *