Google, on its help page says: We calculate an overall rating based on user ratings and a variety of other signals to ensure that the overall score best reflects the quality of the establishment.
Obviously they are not calculating the average the way that most people calculate averages. If, like in the screen shot, you have eight reviews that are all 5-star reviews you would be hard pressed to come up with 4.9 average if you did the maths.
There has been the suggestion that they may be counting other sentiment around the web which is a possibility. But in many cases I have looked at there are no other bad reviews around the web.
What is going on?
When you use reviews for ranking, like Google and many other companies do, you have a problem. Is a business with eight 5-star reviews actually better than a business with 48 5-star reviews and two 3-star reviews? Should the one with eight actually rank higher?
Probably not. Sooner or later the company with only eight reviews is likely going to get some lower reviews. Or so Google assumes and they don’t want to give them more credit than is their due. What is a company like Google to do?
What they always do. Write an algorithm.
Google clearly is weighting 5-star reviews differently if there are less than a certain number and they are probably using an algorithm that is something like a Bayesian average.
What is a Bayesian average you ask? From Wikipedia:
A Bayesian average is a method of estimating the mean of a population consistent with Bayesian interpretation, where instead of estimating the mean strictly from the available data set, other existing information related to that data set may also be incorporated into the calculation in order to minimize the impact of large deviations, or to assert a default value when the data set is small.
In other words, when you have a very small sample (i.e. few reviews), Google is using its large dataset of review stars to model and estimate what your average would be IF you had more reviews. All very Googly, as they say.
The problem is that Google doesn’t really explain all of this and folks fret a great deal. Often for good reason. The Bayesian logic, while integral to Google’s thinking, is very foreign to most of us. Here is a tale from the forums when I explained how Google likely calculated this “average”:
Well that is disheartening. I’ve already had one person accuse me of having a bad review that Google is “hiding” for some reason because it’s so obvious that the math doesn’t work out without a low review being factored in somewhere, I can only imagine that other people investigating my business might think the same. I really wish they would just use actual numbers instead of “assuming” that I’ll get a bad review at some point. I think real life is a lot more useful to customers than predictions!
What can a small business do? Not much, you just need to keep at the project of providing great service AND encouraging your clients to leave reviews. Sooner or later (20, 30, 40 reviews later?) Google will calculate a real average for you.
What should Google do? Be more transparent and tell people how this works. (It’s that time of year when I can ask for presents, no matter how unlikely I am to receive it.)
Article by Mike Blumenthal of Localu.org