How Are Amazon Product Ratings Calculated?

Amazon uses a highly sophisticated algorithm to calculate product ratings and to decide which reviews should be given the most weight. Here’s how it works.

Amazon dominates the ecommerce marketplace. More than half of all purchases made online last year in the US were made using Amazon. And, its Seller Marketplace is growing at an exponential rate adding around a million new sellers every year.

One of the keys to Amazon’s success in this space has been its excellent product reviews system, but how does this amazon rating system work and can you really trust all the reviews that you see?

Amazon Review controversy

Amazon’s approach to calculating the average product rating is a source of constant angst for many sellers and understandably so. It’s hard to overstate the importance of product reviews to Amazon sellers. The vast majority of buyers will at least read some reviews before making a purchasing decision. Various surveys put the number of buyers who read product reviews before purchasing near 90%. It is important to consider the value buyers place on Amazon product reviews. A recent survey found that 84% of people trust an online product review nearly as much as they trust a recommendation from a friend.

At the very least, buyers will take a quick look at the average star rating for each Amazon product and many will go looking specifically for negative reviews just to see what could potentially go wrong with the product. If a product has lots of complaints or a below-average star rating, buyers may decide that the product is not worth their time or money.

The problem, from a seller’s perspective, is that the math of Amazon’s product review algorithm doesn’t always make sense. A quick glance at the forums on Seller Central will confirm this to be true. Take this complaint from one exasperated seller, for example:

“I did my calculation on the review rating. 90% of 5 stars and 5% of 4 stars and 5% of 3 stars give you 4.84, but why Amazon has 4.7? Drive me nuts. It has a big effect [sic] on my sales. Amazon doesn’t know how to do the math!!”

This Amazon seller is not alone. There are many other frustrated sellers convinced that Amazon is under-estimating their scores. So, do the people at Amazon need lessons in calculating a mean average or is there something else going on?

Amazon Product review algorithms

Simply put, Amazon’s product rating calculations can seem off because, in their eyes, not all product reviews are created equal. Until a few years ago Amazon’s method for calculating an average 5-Star rating was extremely simple. They would gather all the reviews and present the rating as a straightforward average that everyone could understand. In 2015, Amazon changed its rating calculation to add more weight to certain reviews.

Amazon now looks at specific factors contained in reviews to determine how much each review affects the product’s overall rating on Amazon. Some of these factors are whether or not the review is a verified purchase, the number of people who say they have found the review helpful, the age of review, and the length and detail contained in the text of the review.

When you think about it, this approach to product ratings makes sense. At the time of the change, Amazon spokeswoman Julie Law said it was designed to better reflect the product as it is – rather than as it has been in the past. For example, if someone were to make changes to their product but these changes make the user experience worse, the new rating system picks this up. In this case, the recent reviews would be worse than the old ones. Under the old system, the average 5-star rating would have been propped up by the older, more favorable product reviews.

Of the change, Law said, “It’s just meant to make things that much more useful so people see things and know it reflects the current product experience.”

The Amazon product rating algorithm can help sellers who have made positive improvements to their products. Perhaps the initial product was not great and garnered a lot of negative reviews. The seller listened to customer complaints and fixed what was wrong. The newer reviews will be better than the old ones and reflect the positive changes. Amazon’s calculation will give more weight to the more recent product reviews, helping the seller to succeed.

Fake reviews and Amazon’s product rating algorithm

In certain areas, fake reviews are quite common. Unscrupulous sellers are buying fake positive reviews for their own pages, and in some cases, paying for fake negative reviews on their competitors’ pages.

Fake reviews pose a massive problem for Amazon. Amazon buyers trust in the review system and it is one of the reasons Amazon is the go-to ecommerce marketplace. Buyers use these reviews to determine whether or not they should purchase an item. If the trust in product reviews begins to erode, it won’t be long before trust in Amazon as a whole starts to erode as well.

This is why Amazon has taken its efforts to fight fake reviews to the next level. Amazon’s rating algorithm will search for tell-tale signs of fakeness. These signs include overly short reviews, many reviews left over a short period of time, or reviews left by people who only leave definitively positive or negative reviews. If at any time a seller suspects that a review is fake, they can contact Amazon and request that action be taken to both remove the review and suspend the reviewer.

People who want to scam the system will always try. This report on the Verge highlights one example in which a seller noticed a spate of 5-star reviews flooding onto his page over the course of a single day. Given that his previous review average was one review per day, he was suspicious of these reviews and reported them to Amazon.

This seller was right, something was amiss. A few days after the reviews were removed from his page, Amazon suspended his account for buying fake reviews. He’d been set up. A competitor had launched a campaign to get him thrown off Amazon by framing him for one of the most serious of Amazon seller crimes.

High-quality vs low-quality reviews

Even among real reviews, some will be more useful than others. For example, a customer who has taken the time to understand a product will produce a more valuable review than someone who has just used it a few times and posted a quick review without giving it too much thought. The difference often lies in the length and detail of the review.

As buyers reading reviews, we tend to do this naturally. We will instinctively have an idea of how much we should respect the opinions stated in a particular review. Either consciously, or subconsciously, we assign a rating to each product review giving it greater or lesser influence on our buying decisions.

Amazon’s algorithms try to do the same thing. They will weigh each review based on various factors. These factors include how many people have said they find the review useful, the age of the review and any written comments. If a product review is listed as a ‘verified purchase’, it will also carry more weight than one which is not.

Unverified purchases are a flaw in Amazon’s product review process. Anyone can go on and leave a review whether they have purchased the product or not which leaves Amazon’s product rating algorithm open to abuse. Sellers have been known to pay people to leave positive reviews on their profiles and bad reviews on their competitors’. There are even sites dedicated to helping people buy product reviews.

Improving your product reviews

For sellers who have taken the time to check Amazon’s math on their product reviews, Amazon’s process can seem vague and unfair. The exact way the rating algorithm works will remain something that only Amazon understands. But, understanding which factors in a product review is more heavily weighted by Amazon’s rating algorithm can help sellers improve their reviews.

The best way to improve product review ratings is to increase the number of reviews received. By encouraging more buyers to leave reviews, sellers will increase the number of high-quality product reviews from verified buyers that appear on a product page.

Although Amazon has strict guidelines that forbid sellers from specifically asking for positive product reviews, sellers are still able to communicate with buyers in a way that makes it more likely that they will receive one.

Automating an email sequence can go a long way to supporting buyers and helping them to better understand the product they have just purchased. Send an email to confirm an order, and a second one when the item has shipped. With this second email, sellers can include extra information, like tips and tricks for how to get the most out of the product purchased. This will help the buyer maximize their user experience and increase the likelihood that they view their purchase in a positive light. Finally, sellers can send an email a few weeks after delivery just to confirm that everything is going smoothly.

By sending a simple email sequence like this, buyer engagement is increased and the chances of receiving a positive review are increased as well.

One of the reasons Amazon remains so successful is that they keep their methods to themselves. To this end, sellers are unlikely to ever fully understand how Amazon calculates product ratings. But by constant monitoring of product reviews, and engaging with buyers, sellers can boost their rating and increasing the number of verified reviews that appear on a listing.

It is not easy, but it is possible!

Want help getting more reviews to improve your Amazon product ratings? Looking for help with Amazon email automation? FeedbackWhiz has email automation and Amazon product review tools that will help make your job as an Amazon seller easier. Try out FeedbackWhiz with a free 30 day trial. Sign up today!