A shopper searches for a phone charger, lands on a single product page, and sees one price with a big "Buy Now" button. Refresh the page a day later, or check from a different account, and the price has shifted by a dollar or two. Sometimes there's a small note: "sold by" a company the shopper has never heard of. Nothing about the page signals that this listing is actually a marketplace, not a store. It's a shared shelf where multiple sellers, and sometimes the platform itself, compete to fill the order.
Behind the scenes, a contest decides that single price. The platform doesn't simply set it.
Large online retailers run two businesses on the same website. In the first, the platform buys inventory from manufacturers at wholesale rates, holds it, and resells it directly as first-party retail, setting the price the way any traditional store would, balancing margin against demand. In the second, independent sellers list their own inventory on the platform's product pages and set their own prices, and there the platform acts as landlord: it hosts the listing, often handles warehousing and shipping, and collects a cut of every sale. Both kinds of listings can appear on the identical page, competing for the same customer click.
When multiple sellers offer the same item, the platform's software decides whose price and "buy" button gets the prime spot. That's commonly called the buy box. Not any single seller makes that call. A scoring system weighs price, shipping speed, seller reliability, and inventory availability instead. Sellers who don't win the featured slot don't disappear. Their offers sit in a secondary "other sellers" list that most shoppers never bother to open.
Winning the buy box, more than setting an appealing price in isolation, matters most. It's the real objective, and it changes how sellers behave.
To win it, many sellers run automated repricing software that watches competitors' listings, adjusts price in near real time, multiple times a minute, and undercuts by the smallest increment the algorithm calculates will work. Pricing becomes a machine-versus-machine race, rather than a judgment call about margin or brand positioning. A seller who refuses to participate simply loses visibility, regardless of product quality or service. Pressure compounds here because the platform earns its commission and fulfillment fees on the transaction no matter which seller wins. So the platform benefits from the price war even as individual sellers' margins get squeezed by it.
That fee structure is the profit engine underneath the whole arrangement.
It charges a referral fee as a percentage of each sale. Storage and fulfillment cost extra if the seller uses the platform's warehouses. Advertising fees apply if the seller pays to appear above organic search results. None of those charges depend on whether the seller actually turns a profit on the item itself. A seller can win the buy box, sell in volume, and still lose money once fees are subtracted. The platform collects steadily on every transaction that crosses its shelf, win or lose for the merchant.
Here, an incentive structure emerges: sellers increasingly compete on advertising spend rather than pure price. Paying for placement is a more controllable lever than an endless repricing war that erodes margin. It also lowers the bar for entry. Because listing costs little upfront, the marketplace side attracts high volumes of sellers, including resellers of gray-market or counterfeit goods, who can undercut authorized sellers precisely because they didn't pay the legitimate acquisition cost. Genuine brand owners face a choice: match those prices, lose the buy box, or pay the platform for advertising to get around the problem. All of those outcomes funnel more revenue back to the platform.
Here's where the incentives get most tangled.
The platform is both referee of a marketplace and a competitor in it, since it also sells as a first-party retailer. A platform can decide to price its own first-party inventory below cost on an item that drives traffic or supports a subscription program, but an independent marketplace seller generally cannot afford to do the same, since sellers lack the platform's other revenue streams to subsidize a loss leader. The same company sets buy box rules and also fields its own listings competing under those rules. Sellers reasonably ask whether the scoring favors the house. Platforms typically respond that first-party and third-party offers are scored identically, but the process itself is not published in detail. So the claim is difficult for an outside seller to verify.
This opacity is central to the arrangement's durability. Sellers who suspect the buy box algorithm disadvantages them have little practical recourse. There is no published formula to appeal to, and no regulator-mandated audit trail. Switching to a competing marketplace means rebuilding reviews, rankings, and customer trust from zero. Meanwhile, the platform has no strong internal incentive to publish the scoring logic. Ambiguity keeps sellers competing harder, on price, on ad spend, or both, without a clear ceiling on what's demanded of them.
Opacity, in short, is a feature of the business model, not an oversight.
What would actually shift the balance is external pressure. The platform has little reason to change it voluntarily. Regulators in some jurisdictions have begun treating large marketplaces as gatekeepers subject to rules against self-preferencing. They're requiring clearer disclosure of when a platform's own retail arm is competing against the sellers it hosts. In some cases, they also restrict a platform from using seller data gathered through the marketplace to inform its first-party buying and pricing decisions. Separately, a platform could restructure itself to exit first-party retail entirely and operate purely as a marketplace, which would remove the conflict of interest at the root rather than policing it after the fact. Some platforms have discussed that option but rarely completed it at scale, since first-party sales remain a reliable revenue base.
For now, the more realistic source of change is competitive. Sellers and brand owners can route around a marketplace's fee structure two ways. They can build their own direct-to-consumer channels, or concentrate volume on marketplaces with more transparent seller terms. Either move pressures the incumbent platform to loosen fees or open its scoring criteria, to keep sellers from leaving. If it happens, the shift will show up gradually in seller behavior before it shows up in policy announcements.
Readers who want to see the mechanism in action can click the "other sellers" or "compare offers" link beneath a product's main price. It usually reveals a spread of competing prices the buy box otherwise hides. Three things would show the shift is real: whether platforms disclose more about buy box scoring under regulatory pressure, whether any major marketplace spins off or scales back its first-party retail business, and whether sellers begin publicly steering customers toward their own websites to escape marketplace fee structures altogether. Any of those would signal that the underlying incentives, not just the sticker price, are starting to move.