Self rating. How about that?

All of us have come across and submitted a star rating to a service or a location. While five star ratings are the most widely used form of rating experience, it has its own disadvantages. The biggest problem is that different people interpret it differently. 3 stars might be “average” to someone while for some others, a great service is a 4 and 5 is supposed to be extraordinary and beyond expectations. The problems of a 5-star rating system are well documented, in particular the phenomenon of a J-curve — users only rate something genuinely and seriously only if they are either super happy(5) or super frustrated(1) and everything else in the middle is almost arbitrary.

Having said that, not all 5-star ratings are unsuccessful. There are implementations that work very well in some contexts. An example that works for me is restaurant ratings on Yelp. Being a foodie, I try to go to a different place every time I want to go eat outside (you can assume that is the case with most Yelp users. They installed the app because they can discover places around them that they can go and have fun at). Over time I established a benchmark that works for me — I would try a restaurant that serves Indian cuisine if it has a 4+ rating. This data is user generated and as a provider of a given service, it’s so hard to request this subjective feedback from a user and show that as a qualitative measure to the rest of the world.

For an average user with no negative or positive emotion after using a service, picking the right star is a task in itself. And I was fascinated when I saw this pattern on this survey I got after I bought my new car.

Survey from AnyPerk

After answering a series of questions, the software filled a value for the rating based on my answers and let me change it if I wanted to. Honestly, I am not sure if it’s effective or influencing my brain in some sense but it definitely is an interesting interaction to explore. I could not use the survey multiple times but I am assuming it would calculate a different rating if my answers were different.