Amazon looks to improve customer-reviews system with machine learning
The e-commerce giant creates a new in-house tool designed to make its customer-reviews setup more up-to-date and helpful.
Amazon is rolling out a big change to its customer reviews system in the US, introducing a new machine-learning platform it developed in-house to surface newer and more helpful reviews.
"The system will learn what reviews are most helpful to customers...and it improves over time," Amazon spokeswoman Julie Law said in an interview. "It's all meant to make customer reviews more useful."
The change, which started Friday, will probably go unnoticed at first, as the e-commerce giant's new platform gradually starts altering the star ratings and top reviews on product pages. The new system will give more weight to newer reviews, reviews from verified Amazon purchasers and those that more customers vote up as being helpful.
A product's 5-star rating, which previously was a pure average of all reviews, will also become weighted using those same criteria, and so may change more often.
There's no word yet on whether this new platform will roll out to other countries.
Customer reviews have been a crucial part of Amazon's websites for over 20 years, with the written reviews and 5-star rating system becoming an important form of accountability and sign of popularity and quality for items buyers often can't touch or test out before purchasing. Because the system has become so important for selling products online, Amazon carefully considers any changes to the ratings system to ensure customers continue to trust it. Additionally, in April the company said it sued a handful of alleged fake reviews websites to fend off a small but potentially threatening amount of phony reviews.
The new platform was something the company looked at "very closely" before instituting, Law said, though she declined to say how long Amazon had been developing it.
"It's just meant to make things that much more useful," Law said, "so people see things and know it reflects the current product experience."
For example, sometimes a company will make small tweaks to a product or address some customer complaints, though this product isn't officially updated or renamed. With the new system, Law said, these small modifications should become more noticeable when shoppers are buying products.