Bain’s latest retail numbers describe a plateau the last decade of capital was built to outrun. Online’s share of American retail has settled at 16.4 percent, well short of the takeover its backers underwrote. That flattening inverts the store’s place on the ledger, recovering as a margin-and-discovery asset what a decade of direct-to-consumer spending wrote off as a cost to shed. The premise going with it held that digital would keep compounding until the store became a liability. The argument here is not that stores survive, but that the capital fled the wrong line.

The plateau shows up first in the growth rate, not the level. US online sales grew 5.4 percent in 2025, by Digital Commerce 360’s reading of Commerce Department figures. That was the fourth year running in single digits, which is not the arithmetic of something still opening a market. A channel that spent the 2010s taking share by the year now takes it by fractions of a point. The takeover thesis needed the curve to keep bending upward, and it flattened.

Bain’s own 2025 forecast concedes the point in the act of sounding bullish. Reported by SGB Media, it expects online sales to grow 10 percent, a rate that impresses only until you notice stores are still growing too, on a far larger base. Weight that base and the gap narrows, because stores still ring up close to five of every six retail dollars, and the smaller rate is bolted to the bigger number. Counted in dollars, the two channels are dividing the market’s growth rather than moving it from one side to the other. A decade of capital assumed the reverse, that the flow ran one way and would not slow.

What looks like a pause is, in the categories that led online, a ceiling. Bain has flagged several of the segments that went digital first, among them diapers, infant formula, skincare and cosmetics, as they approach what it calls the natural limits of online penetration. The mechanism is unsentimental: as a category’s online share climbs, the next customer costs more to win, so the marginal online sale grows dearer exactly as the easy demand runs out. A channel need not reverse to stop compounding; it only has to exhaust its cheap customers, and in the segments that led, it has. Penetration and acquisition cost pull against each other past a point, which is the fact the takeover story left out. Where that point sits differs by category; that it exists no longer does.


The premise now breaking had real money behind it, which is what makes the plateau expensive. Digital-native brands captured roughly 60 percent of global retail growth between 2013 and 2021, and capital read that share as a forecast rather than a phase. Casper and Allbirds went public promising store fleets and category expansion, then pulled back, with Allbirds shuttering locations and handing whole markets to distributors, as CNBC reported. Bain marked the turn in 2022, when its global retail lead Marc-André Kamel wrote that “the fastest phase of growth for digital-first retailers looks likely to be over.” The capital that filed stores under “the past” was mistiming the present. Its error was not backing digital; it was reading an early slope as a destiny instead of a segment filling up.

Bain’s newer work reads the store the other way, as the asset the plateau leaves standing. In research with VusionGroup, it finds three-quarters of retail executives planning large-scale store transformations, and most of them lifting store-technology budgets rather than cutting them. The reframe is not nostalgia; it rests on two things the online channel taxes and the store gives away. A store meets demand a retailer never had to buy at auction, and it converts that demand at a gross margin the shipping-and-returns math of e-commerce seldom clears. Retailers are re-tooling the floor to capture both, with nearly a third expecting the layout itself to tilt toward retail-media and experiential formats. The floor becomes infrastructure: a fulfilment point, a media surface, and a sensor for demand the website never sees.

On a flattening curve, the cheapest customer a retailer can reach is the one already standing on the shop floor.

The strongest objection is that the plateau is an interval between interfaces, not a ceiling. On this reading, agentic checkout, social commerce and AI shopping assistants will open online demand the present category mix cannot capture, the way mobile revived a desktop channel that had also looked mature. If that holds, the compounding resumes, and the capital that fled the store was early rather than wrong. The objection is serious, and it breaks on the mechanism it has to overturn. A saturated category stays saturated whichever interface places the order, because an agent changes who captures the reorder, not whether a household needs a second case of diapers online. Bain frames the coming decade as convergence for the same reason, with legacy chains buying digital and digital natives buying stores, which is not the shape a market takes when one side is about to win.

That leaves retailers holding a decision the plateau has already framed for them. If online keeps compounding at the single-digit rate the data now shows, the retailer who closes stores to fund a channel growing 5 percent a year is defunding its best discovery point to feed its dearest one. The question is no longer whether to hold stores, but what to make them do. Two answers are already visible in today’s coverage: Claire’s re-tooling its floor into a discovery-and-content asset, and Foot Locker renting its stores to Uber as fulfilment nodes in someone else’s app. One keeps the customer the store was built to find; the other hands that customer to a platform and keeps the rent. The plateau does not decide which is right — but it has closed the era in which shedding the store could pass for strategy.