Accurate Rating Systems

In class on Wednesday we discussed how to monitor ratings on websites in order to ensure they are accurate. We talked about how some people can manipulate the ratings to improve their own situation. For example, a restaurant owner may give his business good ratings and give the restaurant across the street bad ratings in order for him to get more customers. The ratings could be wildly inaccurate but a prospective customer on a review website would not be able to know that. In class, several ideas were given for how to prevent against situations like this. My own belief though is that the best way to ensure the ratings are accurate is simply to have a large number of reviews. If the sample size is large enough, the random and inaccurate will not have an effect because the majority of the accurate ratings will outweigh the inaccurate ratings. This would work for product ratings, restaurant ratings, class ratings (TRACE), and many others. However, while the idea is simple, getting a large number of ratings is not.

In class we also discussed how many people actually fill out surveys and rate products and services. A large number of students said they didn’t fill out ratings and surveys and the main reason why was that the time investment was too large. Even the students who said they did fill out ratings said they did so mostly when it was convenient and easy. In my personal experience, I will often fill out a survey with only radio buttons and checkboxes and skip sections where I actually have to write something. I think the best way to get a large number of ratings is to prompt the customer (via email or another means) and to keep the survey or ratings as short as possible. For example, an email asking the customer to simply rate a product on a scale of one to five stars. The customer would not have to remember to go to a website, they would be sent an email to remind them. The simplistic rating would also take the same amount of time to complete as deleting the message would, so the customer would most likely rate the product. This would allow for a large number of ratings to give an accurate overall picture of the quality of the product, restaurant, or service. The customer could then also be given an option to write a review and/or complete a more detailed rating in order to enhance the overall rating. A prospective customer could then view the detailed ratings but would be able to see the average star rating that was taken from a large sample size to determine if the detailed rating is reliable.