How the Buy Box Is Won

Several sellers can offer the same product on the same page. One of them gets the Add to Basket button, and that seller takes the overwhelming majority of the sales on that page. Everybody else is behind a link most shoppers never open.

That button is the buy box, and the share of page views where you held it is what Seller Central calls your buy box percentage.

It is a promise about the delivery, not a reward for the seller. Amazon is choosing which offer it is willing to stand behind for a shopper who does not want to compare merchants, and most shoppers do not.

That framing explains almost every rule underneath it. The winning offer is the one Amazon expects to arrive fastest, in the condition described, with the least chance of a problem afterwards.

Delivery. The single biggest input. Amazon says the fulfilment method itself is not scored and the delivery promise is: the speed, free shipping and free returns. An FBA offer is competing on Amazon's own promise, which is fast and free by construction; a merchant-fulfilled offer is competing on yours, and yours has to be genuinely fast, genuinely free and genuinely reliable to keep up. This is the reason most of the buy box question is still really the FBA or FBM question, even though the method is not the thing being weighed.

Landed price. Item plus delivery, not the item alone. An offer a little dearer with free delivery frequently beats a cheaper one with postage on top.

Seller performance. Late shipments, cancellations, defect rate, response times. These are the metrics on your account health page, and they are the ones that quietly cost you the button before they ever cost you the account.

Stock. An offer that cannot ship is not an offer. Running out does not just stop sales that day, it hands the position to somebody else and you have to earn it back.

Condition. Used offers do not compete with new ones for the same button; they have their own.

Amazon's own name for the button is the Featured Offer, and that is what Seller Central reports it as; buy box is the name sellers use. Until 2026, winning it was a two-step process. First Amazon decided whether a seller was eligible at all, on account-level performance, and only then did it rank the offers of the eligible sellers on the measures above. A seller whose metrics fell below the bar was out of the running on every listing, whatever the price or the delivery.

Starting in July 2026 Amazon removed the eligibility step, rolling the change out store by store through the end of the year, and its announcement put it plainly: every seller's offers are now in the running to be evaluated. Nothing changes in the ranking. Amazon lists the same six things it weighs, total price with delivery, delivery speed, free shipping, free returns, seller performance and stock, and it names four things it does not weigh: customer reviews, star ratings, the Prime badge and the fulfilment method.

So performance did not stop mattering; it stopped being a locked door. An account with a dented late-shipment rate is no longer shut out of the button on every listing, but its offers are ranked with that rate counted against them, so the fix is the fix it always was. What has changed is the competition: more offers on every page are in the running, including offers from sellers who used to be gated out.

The change also produced the forum threads. In the weeks after it reached their stores, brand owners who are the only seller of their own product reported their Featured Offer share falling from 100% to somewhere between a half and three quarters with nothing changed on their side, and Amazon's forum staff had no explanation beyond the general one. That is a reported experience, not a rule, and the useful reading of it is the next section: when the share moves and you did not, look at what Amazon is comparing your offer against, which for a sole seller means the pages where the button was shown to nobody.

Because the other side of the comparison moved. A competitor came back into stock, someone repriced, a seller with faster delivery arrived, or your own stock got low enough that Amazon started splitting the page between offers.

The button can also rotate between comparable offers, so two sellers close on every measure share the page over the day rather than one taking all of it. A percentage that sits in the seventies is often that rather than something wrong.

And on some pages nobody wins it. When Amazon judges the price too high against its own reference, the button disappears for everyone and the page shows a See All Buying Options link instead, which converts far worse for all of you.

Fulfilment method, stock cover, landed price, and the performance metrics. That is the list. Everything else is somebody else's decision.

Which means the practical work is unglamorous: keep the fast-moving lines in stock, keep delivery promises you can meet, price including delivery rather than beside it, and fix defect and late-shipment rates before they matter.

Automatic repricing wins the button by cutting the price against whoever is cheapest, which works and is also how a category ends up unprofitable for everybody in it.

Set a floor that reflects your real costs, including fees, returns and advertising, and let the tool work above it rather than through it. A rule that has no floor eventually finds the price at which you lose money on every sale, and it finds it faster than you will notice.

Then the question changes. Other sellers offering your product is legitimate if their stock is genuine, and the answer is not usually an enforcement report. It is Brand Registry, control of the listing, and being the offer Amazon prefers on every measure that decides the button.

Distribution is the other half. Most unwanted sellers arrive because somebody in your own supply chain sold to them, which is a contract question rather than a marketplace one.

Buy box share is one of the first things we read on a new Amazon account, because a listing that converts and does not hold the button is a different problem from one that holds the button and does not convert, and the fix for one does nothing for the other.

The same shape of question exists elsewhere. Walmart also picks a winning offer, and eBay decides which listing to show rather than which seller, which is why the same catalogue performs differently on each.

If your share is falling and nothing on your side changed, it is worth having somebody read the numbers.