In an effort to make product reviews more useful to its customers, Amazon is making several major changes to the way reviews are displayed and ratings are calculated.
Based on machine learning technology it has developed, the ecommerce giant will weigh more heavily newer reviews and reviews from verified Amazon purhcasers.
Additionally, it will take into consideration upvotes on reviews from other Amazon customers.
Star ratings, which in particular can make or break a product on Amazon, will become more dynamic too, with Amazon placing emphasis on ratings from verified Amazon purchasers.
Initially, the updates to Amazon’s reviews and ratings system is being launched in the United States. Because the updates are driven by machine learning technology, Amazon says that the results should improve over time.
According to Amazon spokeswoman Julie Law, the company doesn’t take changes like these lightly and evaluated the implications “very closely.”
While Law indicated that the changes were designed to help customers better understand the “current product experience,” they are ostensibly also intended to thwart fake reviews, a significant problem that Amazon has grappled with for some time.
Amazon was an innovator with its use of reviews, as well as the way in which it used them on its product page for maximum impact.
However, since anyone can leave a review, the credibility of the whole lot may be in danger.
Since Amazon is such a massive ecommerce player, there is a real incentive for unscrupulous brands and marketers to try to game the system by leaving fake positive reviews on their own products, or by dissing those of competitors.
In fact, there’s a massive industry built up around fake reviews, especially targeting Amazon.
Then there are reviews which just aren’t any use for customers, such as those due to the ‘fanboy effect’, where Xbox and Playstation fans use Amazon’s reviews to argue over who has the best console, or political books, where entrenched views mean that reviews are highly polarised:
There have been calls to ditch unverified reviews, and this move is one way for Amazon to address the credibility issue.
PageRank, EdgeRank and now ReviewRank?
Companies that market and sell online have become increasingly aware of and impacted by algorithms. From PageRank to Facebook’s old EdgeRank, numerous algorithms have a great ability to influence whether a company succeeds on the web or not.
For companies which have products sold on Amazon, Amazon’s new algorithm could be a game changer, for good or bad.
Unfortunately, outside of creating great products that customers want to rave about, it appears companies may have a difficult time optimizing for Amazon’s machine learning system, as in many if not most cases they will have a limited opportunity to encourage an ongoing stream of reviews and ratings from Amazon verified purchasers.