Review sites are underutilized

Recently, review websites sparked my curiosity and I’ve decided a few of my insights on them (don’t want to give away all my ideas). Review sites use crowdsourcing to try a form a more accurate picture of the general public's perception of that company. They generally do this by allowing users to input a comment about the place, and a 1-5 rating of it that is usually displayed as an overall average and a bar chart. This provides value to the customer by allowing them to use the rating to discern the overall picture, while still being able to read up on personal tidbits of what a person’s perception of the company is.

Some of the problems currently with review sites are: the companies are incentivized to pay for reviews to cheat the system in order to increase their personal revenue, users are biased in their reviews, there may not be enough information to accurately judge the company, and reviewers may not know what they are talking about. Fortunately, most of these problems have pretty good solutions waiting to be implemented.

One of the most important things review sites can do to increase their value to customers is to increase the number of reviews. This is due to the law of large numbers that states that the average value of an experiment performed a large number of times will regress towards the expected value as more trials are performed. This means that as the number of reviews goes up, it becomes more likely that the average rating reflects the company’s actual public perception. This will not only drown out the random occurrences, but also will disincentive the companies from trying to pay for reviews because as the number of reviews goes up, the effect of each bought review goes down, thus decreasing the return on investment for buying reviews (assuming the cost of buying reviews stays the same).

To achieve an increase in reviews, one can apply a basic law of economics that states that people respond to incentives. A great example of this is Amazon Vine. Because reviews give a product more credibility, companies are incentivized to give away free products when they are first released in order to encourage users to buy that product in the future. Amazon Vine reviewers are incentivized to write good reviews because they know if they do, they can get free stuff. Thus, the middle man (Amazon Vine), experiences almost no cost to get more reviews that increase the credibility of its products and therefore increases it’s sales. This can be applied pretty easily to review sites through things like gift cards or other rewards for writing an in-depth review. Now to maximize the value created for the review sites, the optimal reinforcement schedule would probably be variable-ratio. Examples of this type of reinforcement are slot machines, which reward people inconsistently while they are usually losing money. This leads to a high, steady response rate, and because it is still somewhat random, it could be influenced by the quality of a review, thus promoting more valuable reviews.

While having lots of reviews is helpful to generalizing a company. Review sites still suffer from reviewer bias and ignorance. Fortunately, statistics has solved both of these problems through the standard score, which represents the number of standard deviations a data point lies above or below the mean. This can be applied to review sites where Z = (x - μ)/σ where Z is the standard score, x is the raw user score for the company, μ is the mean of the user’s scores, and σ is the standard deviation of the user’s scores. The reason this is helpful is because if a user gives every place he visits a 5 star review, it does not really add any value in comparing companies and therefore should not be taken into account for the company's rating. But, if a user gives out mostly 3 star reviews, a few 4 stars, and sparingly gives out 5 star reviews, then we know that the businesses with a 5 star review must be pretty good. Not only does the standard score help compare companies, but it makes it much harder to fake reviews because if a reviewer’s reviews do not follow the normal bell curve, then they are most likely a fake.

These are but a few my ideas on the underutilized value review sites can offer. Hopefully, they will pick up on these and improve themselves, or hire me for an internship/job to do it for them :).