Amazon is trialing a complex new approach to product ratings in a trial feature raises concerns.
E-commerce operations like the behemoth Amazon understand the challenge of upholding genuine reviews. Sellers often resort to questionable tactics to sway product ratings in their favor, sometimes resorting to outright fraudulent practices to boost sales.
Maintaining the integrity of reviews is paramount, just as accurately conveying them to consumers is. Amazon is now experimenting with the presentation of its product ratings, but it’s uncertain whether this is a move in the right direction.
How star ratings work
Traditionally, Amazon has displayed product ratings in search results using a five-point scale, depicted by yellow stars adjacent to the product image. The rating is made up of a weighted average of all star ratings received by the product.
Let’s say an item gets seven five-star ratings and an equal number of one-star ratings, the displayed average would roughly settle around three stars, considering a total of 14 ratings.
According to the company’s support documentation, it’s not just straightforward averages. They make use of a machine-learned model that factors in variables such as recency and verified purchase status. In practice, the presented value closely resembles a basic average of all the ratings.
Amazon’s AI blunder
However, Android Police recently picked up Amazon’s trial of a novel rating system on both the website and app.
This new system replaces the weighted average star rating with a single yellow star, accompanied by the percentage of five-star ratings in relation to the overall ratings.
For the same example mentioned earlier, an item with seven five-star ratings and seven one-star ratings would display a search page rating of “50% 5-star,” followed by the preceding weighted average rating of three.
And how’s that going for you?
This new system has raised concerns and for obvious reasons. It evaluates products favorably solely based on the percentage of five-star reviews, disregarding whether other ratings lean positively or skew toward negativity.
This approach could also make it easier for sellers to strategically collect sufficient five-star reviews to offset genuinely negative ones. In contrast, a weighted average rating is more resistant to manipulation through paid or counterfeit positive reviews, which are often difficult to detect.
From a user experience perspective, the new rating lacks the immediate clarity of the familiar five yellow stars. Moreover, the mention of “5-star” and the use of a single yellow star in the new format might misguide new users.
In Amazon’s defense, users can still access a breakdown of rating percentages on the product page if the new rating is too vague.
This experimental feature was seen in the Amazon India mobile app, the Amazon Germany website, and the global website accessed from Germany. This does look like a trial of a new feature and there’s no confirmation that it will roll out to the larger public anytime soon.
Share this article
About the author
Marce has contributed tech to various prominent publications since 2018, offering a transparent perspective into the tech industry and its effects on its users. She now spends her time developing insightful content for industry players. You know, when she's not gaming or geeking out about the latest fad.
