Amazon uses artificial intelligence to improve review/rating system

Discussion
Jun 23, 2015

Reviews and ratings are an important part of consumers’ shopping journeys, but false reviews can upset the trust shoppers place in retail sites. Amazon.com has introduced a new platform it developed in-house that promises to make the reviews and ratings on its site more useful for its millions of customers.

Amazon spokesperson Julie Law wrote in an email to InformationWeek, "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."

The new system will use a variety of criteria to determine the relative rank of reviews on the site. These include date (newer reviews are weighted more heavily than older ones), confirmed purchase of a product and, most importantly, whether others found it useful.

Amazon review

Source: amazon.com

Back in April, Amazon sued operators of several websites for writing fake reviews for profits on the site. The suit alleged sites including buyamazonreviews.com, bayreviews.net and buyreviewsnow.com engaged in trademark violations, unfair competition and deceptive acts.

Do you expect Amazon’s new platform will provide more useful reviews and ratings for its customers? How will Amazon’s new system affect its business?

Braintrust
"So Amazon, here’s a gift: reviews from people who have a larger span of time between purchase date and review date are far more valuable than people who open a box and write a review without spending enough time using what they bought. Any company can claim to employ "AI" and "machine learning," but execution and knowledge of users is what makes it valuable."
"Reviews are clearly critical in influencing product purchase. What I have noticed is how frequently positive reviews seem to be pushed to the top. I would like to believe that Amazon’s investment in this area will be with a view to increasing the authenticity of the reviews and to provide a balanced perspective, as opposed to optimizing their display to maximize sales."

Join the Discussion!

12 Comments on "Amazon uses artificial intelligence to improve review/rating system"

Notify of

Sort by:   newest | oldest | most voted
Keith Anderson
BrainTrust

It’s great to see Amazon continuously innovating to improve the usefulness and integrity of its ratings and reviews.

More shoppers visit Amazon to research products than any other destination, and product ratings and reviews have a significant influence on sales online and offline.

Unfortunately, unscrupulous sellers and brands have tried to game the system.

This likely isn’t a silver bullet, but it’s good to see Amazon investing continuously in setting the standard for credibility.

Max Goldberg
BrainTrust

Consumers find reviews to be valuable, so anything Amazon and other retailers can do to keep them authentic is important. Amazon wants to avoid being stung by fake reviews. Yelp has been dogged by this criticism over the past few years and it has damaged the site’s reputation.

Adrian Weidmann
BrainTrust

Customer reviews and insights are a valued resource for online shoppers. Amazon obviously recognizes its value and is investing resources to improve this aspect of its online advantage. Brands and retailers also recognize the value of these reviews and inevitably this leads to artificially positive reviews and abuse. Since customer reviews are such a valued differentiator and a competitive advantage for Amazon it is in their interest to enhance and protect this valued initiative.

Phil Wells
Guest
Phil Wells
2 years 3 months ago

All this assumes that Amazon treats all reviews fairly and doesn’t use the new algorithms to bury inconvenient reviews. Amazon does not take prisoners in their efforts to grab market share.

Naomi K. Shapiro
Guest
Naomi K. Shapiro
2 years 3 months ago

It is being proven again and again that customers rate other customers’ reviews as one of the highest factors in their decision making. Me too. I see where the new platform will provide more timeliness, just like many of Amazon’s predecessor algorithms, which Amazon is nonpariel in creating, but I don’t see where/how it eliminates the issue of fake reviews.

Ken Lonyai
BrainTrust

As described here, this is quite unimpressive. It’s taken them what, 15 years to start thinking about this? Why is a newer review necessarily better than an old review? There are much more useful criteria. So Amazon, here’s a gift: reviews from people who have a larger span of time between purchase date and review date are far more valuable than people who open a box and write a review without spending enough time using what they bought. Any company can claim to employ “AI” and “machine learning,” but execution and knowledge of users is what makes it valuable.

Lee Kent
BrainTrust

Definitely a step in the right direction. This is what the consumer wants, but the big question remains: will it work?

Gotta hand it to Amazon … And that’s my two cents.

Kenneth Leung
BrainTrust

Many online shoppers including myself have built in a “skeptic filter” to discount reviews that are too high or too low or based on what we perceive to be silly reasons. Well articulated, newer reviewers from buyers always help raise the votes in my book. Glad to see Amazon putting more tech behind it to make the reviews easier to go through. Don’t think I will ever take off the filter when it comes to peer reviews though….

Shep Hyken
BrainTrust

Reviews and ratings from customers are more important than ever. It’s word-of-mouth endorsements that help sell the books. Anything to ensure that the reviews are legit will only help the buyer’s confidence.

Gajendra Ratnavel
BrainTrust

Absolutely. This type of technology has already been in use for some time on sites like Yelp and it has helped. Fraud in reviews needs to be controlled. In fact, as the online world matures, the reviews (which is online reputation) becomes a little like currency. There is a really good TED Talk on this subject.

gordon arnold
Guest

Artificial Intelligence programming, no matter how broad and complex, always gives the subject of the inquirer a predetermined choice to proceed to a predetermined end. The subject is given a set of values which may or may not be accurate or in tune with their opinion(s). This problem will exponentially dilute or falsify the results even with the best of correction curves applied.

Finding and hiring expert marketing people and statisticians would be a better plan, or at least a start with some real intelligence.

Alexander Rink
BrainTrust
2 years 3 months ago

Reviews are clearly critical in influencing product purchase. What I have noticed is how frequently positive reviews seem to be pushed to the top. I would like to believe that Amazon’s investment in this area will be with a view to increasing the authenticity of the reviews and to provide a balanced perspective, as opposed to optimizing their display to maximize sales.

wpDiscuz
Braintrust
"So Amazon, here’s a gift: reviews from people who have a larger span of time between purchase date and review date are far more valuable than people who open a box and write a review without spending enough time using what they bought. Any company can claim to employ "AI" and "machine learning," but execution and knowledge of users is what makes it valuable."
"Reviews are clearly critical in influencing product purchase. What I have noticed is how frequently positive reviews seem to be pushed to the top. I would like to believe that Amazon’s investment in this area will be with a view to increasing the authenticity of the reviews and to provide a balanced perspective, as opposed to optimizing their display to maximize sales."

Take Our Instant Poll

Will having more accurate reviews and ratings help or hurt Amazon’s sales?

View Results

Loading ... Loading ...