
Amazon is tweaking how it displays a product's rating and reviews - but will Amazon's new system make it easier for people to game the system?
Geekwire says that instead of using an unweighted average to calculate ratings, Amazon will use a new "machine-learning algorithm." The idea is to make newer ratings count more than older ratings, for example.
An Amazon spokesperson confirmed the changes, telling the online publication, "The enhanced system will use a machine-learned model to give more weight to newer, more helpful reviews from Amazon customers. The system will continue to learn which reviews are most helpful to customers and improve the experience over time."
In addition to changing the numeric rating, Amazon will also use an algorithm to determine the order of customer's product reviews.
The manipulation of seller and product ratings is as old as ecommerce - how many times have eBay sellers tried to boost their feedback, for example? And how many times have sellers tried to harm their competitors' ratings?
Another problem occurs when companies give bloggers free products to review, with the assumption of quid pro quo. Some bloggers proactively ask sellers for free products to review, as we discussed in this
January blog post.
Seemingly most sellers would have less motivation for gaming Amazon product reviews than eBay sellers have for gaming seller feedback, since all sellers are associated with a particular product's ratings on Amazon - though if the seller is the manufacturer or brand owner, that's a different story.
eBay is working on its own product reviews as it attempts to construct a product catalog - see this
recent post, "Where Is eBay Hiding Product Reviews."
Amazon ratings and reviews are incredibly helpful - do you think it should toy with changing its current system? What changes, if any, would you make?