Amazon Now Bills Per Rating, and a 1-Star Probably Costs the Same
Amazon made Review Requests, its ad format that asks verified buyers to rate products, available through the Amazon Ads API in US beta this week, and the advertiser "pays only when a shopper leaves a rating or review." Nothing in the note ties price to the stars. A 1-star rating seems likely to cost the same as a 5-star one, so the number to track is cost per net-positive rating.
That sounds like a small accounting detail. I don't think it is. It changes which products should get this budget at all, and I suspect a lot of sellers are going to point it at exactly the wrong ASINs.
What Amazon actually shipped (and what it left blank)
The format was announced on September 29 with an open beta in late October. It shows native review invites to people who recently bought your product, on what Amazon calls its "highest-traffic pages." Shoppers can leave a one-click star rating or add a written review with photos or video. Eligibility is limited to products with fewer than 1,000 total ratings and reviews, and campaigns pause automatically when a product hits that ceiling.
Amazon's only performance claim is that products in a closed beta got three times as many ratings and reviews per week. That came from internal data covering June 3 to 21, 2026. Nineteen days, no disclosed sample size, no control group mentioned. Take it as a direction, not a forecast.
The API release is what makes this real for anyone running bids programmatically. According to PPC Land's breakdown of the release note, you can now create campaigns, pick products, and set a bid and daily budget through the API. What you can't do yet: pull review-specific metrics through the reporting API. Amazon says those "will follow in a later beta update," with no date. So for now, your bidding scripts can spend money on this format but can't see what they bought. Which is a strange way to launch an outcome-billed product, honestly.
Also left blank: what the bid actually represents, whether a star rating and a written review get charged separately, and any minimum bid or indicative price. If you were hoping for a rate card, there isn't one.
Why the 1-star question matters more than the price
My read on the billing goes like this. The FTC's 2024 rule bars buying positive or negative reviews. Amazon's own policies prohibit review gating, which is asking only happy customers to review. A format that charged more for 5-star ratings, or refunded 1-star ones, would walk straight into both problems. So the billing pretty much has to be sentiment-neutral. You're paying for an invitation that converted, and the buyer decides what goes in it.
Think of it like paying a pollster per completed survey. The pollster gets paid whether the respondent loves you or hates you. If your product is mediocre, you are literally paying to accelerate the evidence that it's mediocre.
So the useful metric is cost per net-positive rating, not cost per rating. Some quick illustrative math (my numbers, not Amazon's): say a rating costs you $2.50 in bids. On a product where 90% of new ratings land at 4 or 5 stars, you're paying roughly $2.78 per positive rating, and the occasional 1-star barely moves your average. On a product where only 70% are positive, it's about $3.57 per positive rating, and every third or fourth paid rating is actively dragging the average down. Same bid, very different purchase.
And that's before the average itself becomes the problem.
Buying more ratings for a 3.9-star product mostly buys you a more confident 3.9.
The first five reviews do most of the work
This is the part that surprised me when I went back to the research. Northwestern's Spiegel Research Center found that a product with five reviews has a purchase likelihood 270% higher than one with none, and that the benefit drops off quickly after those first five. The same study found purchase likelihood typically peaks between 4.0 and 4.7 stars, then declines as ratings approach 5.0. Verified-buyer badges improved purchase odds by 15%.
Now look at Amazon's eligibility window again. It runs all the way up to 999 ratings and reviews. A product sitting at 400 ratings is technically eligible, but the conversion math from the Spiegel data suggests the marginal rating there is worth a tiny fraction of what rating number three was worth. Amazon gets paid the same either way, of course.
In fairness to Amazon, there are reasons to want volume past five. Recency matters to shoppers, a few hundred ratings looks more credible than 40 in a competitive category, and a big base makes your average harder for one angry customer to dent. But those are softer benefits, and I'd want the reporting API before paying for them at scale.
Vine and the free button are your benchmark
You already have two cheaper ways to get reviews, so price Review Requests against them rather than in a vacuum.
The free one is Seller Central's Request a Review button. As SupplyKick CEO Michael Slate laid out, it sends a standardized email asking for a product review and seller feedback, once per order, 5 to 30 days after delivery. Amazon also requests reviews on sellers' behalf at no cost. If your team isn't clicking that button (or automating it) on every eligible order, start there. Paying for the ad version while skipping the free version is a bit like buying branded search clicks before fixing your title tags.
The paid benchmark is Vine. Per PPC Land's Vine explainer, Amazon's US page lists enrollment at $0 for up to 2 units, $75 for 3 to 10, and $200 for 11 to 30, charged per parent ASIN. At the top tier, that works out to under $7 per review in fees, plus the cost of the units you give away. There are also unconfirmed trade-press reports of higher fees on some ASINs, so check what Seller Central actually shows you. Vine only works for products with fewer than 30 reviews, and incentivized reviews tend to skew higher: the same explainer cites ReviewMeta finding a 0.38-star gap in favor of incentivized products.
That gives you a rough ladder. Under 30 reviews: Vine plus the free button, probably. Between roughly 30 and a few hundred: this is where Review Requests has a real job to do, because Vine is closed to you and the free button has a ceiling. Past that, you're likely paying for credibility more than conversion.
How I'd run the first test on Amazon review ads
Personally, I wouldn't put this on every eligible ASIN. The filter I'd use for a first test:
- Star floor: only products already averaging 4.3 or higher on their existing ratings. Below that, you're paying to confirm a problem. Fix the product or the listing first.
- Review band: roughly 30 to 300 total ratings. Vine covers the bottom, diminishing returns cover the top.
- Complaint check: skip anything with a recurring complaint pattern in recent reviews. Unhappy buyers are often the most motivated to respond to an invite.
- Baseline first: log new ratings per week and the 4-to-5-star share for two weeks before spending, since the reporting API can't do it for you yet.
Then set a kill threshold before you start. Mine would be something like: if cost per net-positive rating comes in above about $7 (the Vine fee benchmark, before product costs) and the product's average hasn't held steady, pause it. Your margin structure might justify a higher number, and that's fine, but pick it in advance so you aren't rationalizing after the fact.
One more thing to watch: PPC Land notes the release note doesn't say whether these reviews will be labeled on the product detail page. If they aren't, the reviews blend in with organic ones, which is good for you and probably a question regulators will ask eventually.
Where this lands for sellers by spring
Amazon's ad business is under pressure to find new things to sell. Q3 ad revenue grew 11% to $19.64 billion, down from 26% in Q2, and outcome-billed formats like this one are an easy pitch. We've seen the same move elsewhere, like Shopify buying Meta ads by the sale, and Amazon's auction pricing has its own credibility issues, as the FTC's Black Friday surcharge findings showed.
My prediction: when review metrics finally hit the reporting API, I'd bet at least half of advertisers who tested broadly find their cost per net-positive rating above the Vine benchmark on most ASINs, and quietly narrow it to a short list. I could be wrong if Amazon prices this aggressively to drive adoption, which it sometimes does early on.
I don't think Review Requests is a bad product. It might be a very good one for a specific kind of listing. It's just one where the product does the selling, and the ad only decides how fast everyone finds out.