Consumer reviews shift toward evidence as AI changes online shopping
A survey this year cited by Forbes found that 46% of consumers are suspicious of reviews that appear to be AI-written, while 82% said they had encountered fake reviews in the previous year. For shoppers, the emerging question may be less about finding another rating and more about understanding the evidence behind it.
A shopper searching for a mattress, phone, dog food subscription, or kitchen appliance can encounter numerous five-star ratings and a long succession of "Best Of" lists that appear remarkably similar. The abundance of information may offer convenience while also making it harder to determine which opinions come from actual product experience.
Platforms seem to be responding through their own systems. According to MSN, Google continues to update its search-spam systems to identify emerging forms of search abuse. Meanwhile, Amazon said that its systems blocked hundreds of millions of suspected fake reviews in 2025. Amazon also said its legal actions that year helped shut down more than 100 websites involved in attempts to facilitate fake reviews and scams targeting its store. These efforts seem to suggest that removing questionable material is becoming an important part of maintaining useful digital marketplaces, while the format of the review itself may also warrant greater scrutiny.
That shift may raise a broader question about the information consumers receive before making a purchase. The online environment can contain an enormous volume of opinions, yet volume alone may offer limited insight into whether a conclusion rests on firsthand experience, consistent criteria, or verifiable testing. As regulatory enforcement and platform safeguards continue to evolve, the next stage of product reviewing may depend increasingly on the quality and transparency of the methodology behind the recommendation.
Blake Harrison, co-founder and CEO of Consumer Rating , a consumer product review platform, sees that change as an opportunity to reconsider how product information is produced. His view is that credible reviews require people who can physically examine products, apply consistent criteria, and explain how they reached their conclusions. "People want real human beings actually testing products," Harrison says. He says that for Consumer Rating, that means an editorial team works separately from commercial functions, products are obtained for hands-on testing, category criteria are established before individual products are assessed, and research, customer feedback, and firsthand experience are considered together.
That methodology can also become more technical as review categories mature. Harrison points to mattresses as a useful example."Instead of describing a mattress simply as 'firm,' reviewers could help illuminate what the actual differences are using data, tables, and comparisons," he says Clear affiliate disclosures and candid discussion of limitations can further help consumers understand the context surrounding a recommendation. "The future of product reviews is going to be less about how many reviews there are and more about how credible the methodology behind them is," Harrison says.
The economic implications extend beyond an individual purchase. When product information is difficult to assess, consumers may spend additional time researching, return unsuitable products, or become more hesitant about unfamiliar purchases. A review that documents its methods, acknowledges tradeoffs, and demonstrates relevant testing can give shoppers information they can apply to their own priorities. Harrison also sees a possible evolution in the relationship between AI and human expertise. "AI may increasingly help consumers narrow their initial choices, while firsthand reviews could provide the evidence needed to validate those choices," he says
The result may be a review ecosystem where depth becomes as important as scale. More product testing, stronger multimedia evidence, transparent methodology, and specialized evaluation criteria could give consumers additional ways to judge recommendations for themselves. Harrison envisions a future in which thousands of product categories could be assessed against consistent standards, expanding the amount of specialized information available to shoppers.
As digital commerce continues to develop, trust may increasingly depend on the connection between a recommendation and the evidence supporting it. Ratings, summaries, and automated recommendations can remain useful starting points, while transparent testing may give consumers a stronger basis for deciding which information deserves their confidence.
For reviewers, retailers, platforms, and shoppers alike, the emerging challenge is therefore less about producing more opinions and more about making product information easier to verify. The next generation of online reviews may be defined by how clearly they show consumers the work behind the conclusion.
