How Trustpilot Scoring Works

Trustpilot is a wide-spread consumer review platform. Founded in 2007, they have over 1m reviews posted every month from consumers. Reviewers share their experiences with a company, giving a score, from one to five and a free-text comment.

Companies proudly display the scores on emails, websites, and other mediums as a form of social proof. As such, it’s important to understand how Trustpilot works and how you can influence the scores.

How is Trustpilot’s TrustScore calculated?

TrustScore is a measure of overall satisfaction, and can be seen in a similar light to Net Promoter Score, NPS. Unlike NPS, the score is bounded from 1 to 5 stars.

Trustpilot are relatively open in terms of how the TrustScore is calculated, saying there are three factors that contribute towards a company’s TrustScore:

  • Recency of reviews
  • Frequency of reviews
  • Bayesian Averaging.

Let’s now explore each aspect and how it affects your businesses score.

Bayesian Averaging

The most important statement is that they use Bayesian averaging - a technique not uncommon for rating products, forum posts, and in this case - companies.

Wikipedia defines Bayesian averaging as: “A Bayesian average is a method of estimating the mean of a population using outside information, especially a pre-existing belief, that is factored into the calculation.”

Bayesian Averaging really shines where there are few reviews or samples, as it starts with an initial belief. TrustPilot even state their initial belief, by saying that businesses start with a review score of 7 reviews, 3.5 - a very neutral view of a business.

To understand this, let’s think about a new business whose first review is 5 stars. A simple, frequentist average will simply take the score divided by the number of reviews, giving the business a ranking of 5 stars despite only having one view.

Bayesian Averaging implies that given a 5 star review, we’ll update the prior belief (7 reviews of 3.5 stars) to reflect this new review. As a result, the TrustScore will increase, but will not go straight to 5 stars - the company will need to work harder than that.

Bayesian averaging results in a much smoother movement over time, with scores not whipsawing between 1 and 5 stars. It also means that for a business’ score to go up (or down), there needs to be a consistent period of good (or bad) reviews as scores aren’t affected largely by one off reviews.

When time matters - Exponential Decay

So far, we’ve addressed Bayesian Averaging, but now we turn out turn our attention to the next points:

  • Frequency
  • Recency

Trustpilot say that the TrustScore that newer reviews have more weight than older reviews.

They likely do this using a Expontential Decay function. Exponential Decay decreases a score proportial to its current value. The result of this, is that recent reviews are weighted more than older reviews. Evan Miller nicely explains the Expontential Decay function in relation to Bayesian Average in his post here.

Example Exponential Decay of Rating

In the above, we’ve illustrated how a Exponential Decay function may work. The slope of the curve can be changed, and we can’t be sure of Trustpilot’s exact methods.

What this means for your business

For businesses, wanting to improve or even simply maintain their score need to work hard to continually get frequent reviews to maintain recency in their reviews.

Businesses who are undergoing transformation and are actively working to improve their customer service will find that their score changes faster. Of course, the reverse is true, a downturn in service will likely lead to a decrease in TrustScore.

Of course, TrustScore is not a sole metric, and it should be considered with your internal Voice of Customer program. This also has an added benefit for businesses, in that recent reviews have more of an effect on your current score than old ones.

Outside of this, there is a lot we can learn from a mathematical persepective and rating of your products/services. If you’re looking at implementing ratings for your e-commerce store or business, get in contact with us and we’ll be happy to help.