Buy It Again: The hidden owners behind Amazon's best-selling brands

In August 2024, a buyer of Beckham Hotel Collection pillows left a one-star review on Amazon titled “If this is your second time don’t buy.” The full review reads:

The reviewer posted a photo of both tags. The old one reads “Gel Pillow”. The new one reads “Down Alternative Pillow” and, a little further down, “Made for Cafe Casa INC.”

The pillow in question is still sold under the same listing. This week it ranks as the #1 Best Seller in Bed Pillows, with "80K+ bought in past month" and 4.3 stars from 264,682 ratings.

Since May 2024 at least 97 reviewers who rebought the pillows have written that the pillows they received the second time were not the same.

Cafe Casa, Inc. was one of about 240 companies involved in the 2024 bankruptcy of Thrasio.

Thrasio was the best-funded of the Amazon aggregators, the firms that spent five years buying up Amazon brands. It raised $3.4 billion in 2021 alone. It bought more than 200 brands and sold over 24,000 different products a year. Its finance chief, Josh Burke, said that "the vast majority of Thrasio's brands sell their products exclusively on Amazon." More than 100 other aggregators ran the same play.

These brands range from pet odor sprays and mattress protectors to car vacuums, hammocks, trekking poles, pet brushes, insect traps and kettlebells. Ten of the flagship listings I tracked carry 1,223,584 ratings between them today.

The Amazon aggregators may be the most overt case of financial extraction I have covered to date. The financial opportunity exists because Amazon’s platform lacks transparency. A listing’s reviews and ratings are retained through new owners and product changes. The seller name on the page can be any storefront name, and nowhere on the listing does it state the company that actually owns the brand.

Most people who bought from Thrasio will never have heard of it. Today’s article covers how the aggregators used that lack of transparency at a mind-boggling scale, and how to protect yourself if you do choose to shop on Amazon.

I searched the Beckham listing's reviews for "not the same," "used to," "changed" and "old ones," and read every review the searches returned. I counted a review only when the writer said they had bought the pillow before and that the new one was different. Since May 2024, 97 reviews meet that criterion. Amazon's search caps results, so the true count is likely higher.

Several of these buyers compared the two pillows directly. One weighed both: "The old are 14 ounces. The new weigh in at 13 ounces." Another read the manufacture dates and found that pillows "manufacturered [sic] in 9/2024" felt completely different from "the ones manufactured in 5/2024."

To check whether this happens to every pillow, I ran the same searches on Utopia Bedding's hotel pillow. As far as I can tell, no aggregator has owned this brand. Since 2021, 17 Utopia buyers have reported the same thing, including one whose reorder "came with no tags at all." Relative to overall review volume, Beckham draws these complaints at two to three times Utopia's rate. So the product under any Amazon listing can change while the listing, and its reviews, stay the same.

Star ratings and reviews hide the change. Amazon pools every rating since the listing opened. It says its rating model considers how recent each rating is, without saying by how much. The Beckham listing already had 131,642 ratings by August 2021. In October 2026, a verified buyer who still has Beckham pillows from four years ago wrote that "all reviews from the old construction should be removed because the product is no longer the same."

Amazon's rules for sellers cover exactly this case:

In practice, the seller decides what counts as a new version. I found no sign that Amazon checked any of the listings in this piece, and each one kept its ratings through the changes described below.

If we examine the incentives for a moment, my read is that Amazon has little reason to enforce its own policy. It takes a cut of every sale, often more than half of a typical seller’s revenue once fees and advertising are counted. A listing with years of good ratings outsells a new page with none. Enforcing its own rule would mean removing the buy buttons from the pages that sell best.