The labels that built their names by refusing Amazon are now taking their steadiest growth from it. Allbirds, the wool-shoe company that once treated its own website as a moat, now counts the marketplace among the channels it is leaning on hardest. The switch is a bet on demand: Amazon’s search bar surfaces ready buyers more reliably than the owned channels these brands spent a decade cultivating. In a soft year, reliability is what a brand pays for. A decade ago the reverse was doctrine, and Bombas co-founder Randy Goldberg stated it in 2019: the focus was “Bombas.com, building our brand through our own channels, controlling our story, and importantly, our data.” It reads now like a period piece.

The direct-to-consumer thesis was, at bottom, an argument about who owns the customer. Warby Parker and Everlane launched in 2010; Casper and Glossier followed in 2014. All four went online to cut the department store and the marketplace out of the sale, and to keep the relationship and the first-party data those middlemen had always controlled. Amazon was the arch-middleman: it stands between the brand and the buyer, keeps the purchase data, and ranks the brand beside cheaper substitutes, including its own. To sell there was to rent reach and hand back the customer, the one asset the model existed to hold. So most of the class stayed off it, on principle and on arithmetic alike.

The principle did not change; the cost of honoring it did. Allbirds was already struggling, with revenue down 21.2% year over year and net losses widening, in the quarter before it started selling on Amazon in late 2023. It has since gone further, closing its remaining full-price US stores to lean on wholesale and marketplaces. A brand in that state does not go to Amazon for margin, and Allbirds did not pretend otherwise. It went for demand, its chief executive telling Inc. that “a majority of searches for products online start with Amazon, so we want to show up there.” Allbirds projected the channel would supply about a tenth of its sales while improving margins, a striking forecast from a business founded to prove the owned site was enough.

Allbirds is not the only convert; the shift reaches past footwear into beauty. Glossier, which for years sold nowhere but its own site, now reaches Amazon too, though only through an authorized third-party reseller, Front Row Group, not a direct listing of its own. This is a softer version of Allbirds’ move, but the same concession stands: a category direct-to-consumer was supposed to own outright now runs through Amazon anyway. The common thread is where discovery starts: a 2023 Jungle Scout survey found that 57% of US shoppers begin product searches on Amazon, more than begin on any search engine. An owned website works only when the brand summons the shopper first, through paid social, email, or habit; Amazon skips that step, because the buyer is already inside it. In a cautious year the second path is the cheaper bet, because the brand pays to close demand that already exists rather than to invent it.

The owned channel did not only lose the discovery race; its acquisition math inverted. SimplicityDX calculated that the average online merchant lost roughly $29 acquiring each new customer by 2022, up from $9 in 2013, as ad costs and returns climbed. One driver was no secret: Apple’s 2021 tracking limits stripped out the targeting that had made buying strangers on social media cheap. A business that loses money on the first order survives only on repeat purchases it may never earn, which is the corner the DTC class built for itself. Amazon did not fix that math so much as let brands step around it. The shopper waiting there has already resolved to buy; the brand competes to be the choice, not to manufacture the wish.


The marketplace these brands were built to defy did not defeat their owned channels — it waited for those channels to price themselves out, then sold the brands back the demand they could no longer afford to make.

The strongest case against calling this growth is that Amazon is a strip-mine, not an engine. Stack referral fees, fulfillment, storage, and the advertising now required to be seen, and Amazon now pockets more than half of a typical seller’s revenue. Much of that cut is advertising, and it is not optional in practice: the $68 billion Amazon booked in ad revenue in 2025 is, in large part, brands paying to be seen on the platform they came to for its reach. The brand also gives back the customer data that was the founding point of the model. So the thesis fails under one condition: if Amazon sales merely cannibalize owned-channel sales at that take, and the shopper never returns to the brand’s own site, the marketplace is not an engine but a slow liquidation of brand equity booked as revenue. The objection is right about the price and wrong about the alternative, because a brand whose owned channel has stopped growing at any tolerable cost is not weighing Amazon’s margin against a fat direct one; it is weighing it against no growth.

This is the same position we described earlier today, seen from the other side. We reported that Amazon now rents its shopping AI to rival retailers to run on their own sites; the brands selling on Amazon.com are buying the unbundled version of the same thing. Both are paying Amazon for proximity to demand: one licenses the intelligence, the other rents the shelf. In each case Amazon is paid whether or not it makes the sale, the outcome it has engineered over two decades. The DTC decade taught a lesson it did not intend: owning the customer relationship and owning the demand are separate assets, and a brand can hold the first while still having to buy the second.

If Amazon’s surface keeps outperforming owned demand, the choice facing these brands narrows to one they will not enjoy naming. They can be a brand that sells on Amazon, present on the shelf and legible to the shopper, or a supplier that Amazon sells, one interchangeable line in a category the marketplace ranks and, in time, copies. What separates the two is whether the brand still means anything to the buyer once the search bar, not the label, is where the decision gets made. Allbirds is betting it can take Amazon’s demand now and rebuild its own later, which for a company shutting its stores may be the only bet on the table. What it cannot assume is that the demand it is renting will still recognize its name on the day it tries to stop paying. The DTC decade was meant to end with brands that owned their customers; for some, it is ending with brands that lease them back.