Retail’s front door is quietly being moved by a machine, and plenty of big chains appear to be staring at the old entrance. Bain’s latest work lands on an uncomfortable point: owning market share does not guarantee a retailer gets named when an AI assistant answers a shopper’s question. That ought to rattle management teams more than the usual holiday-season chatter.
Discovery Has Left the Old Search Aisle
Bain tested more than 4,400 shopping prompts across four retail verticals and ran them more than 300,000 times through widely used large language models, using Profound data. The result is not subtle. Category specialists frequently show up in AI answers at a rate above what their underlying market share would suggest, while broad retailers do not automatically receive the same treatment.
That makes sense once you stop viewing AI as a glorified search box. A shopper does not always ask for “retailers.” They ask for the best gear for a particular trip, an item for a particular room, or a gift for a particular person. Specialists have a cleaner story to tell. General merchants can have massive scale, but scale gets mushy when the question is specific and the answer needs to be crisp.
For investors, this is a nasty little shift in the competitive map. The old winners had distribution, brand recognition, store networks, paid search budgets, and years of customer habit behind them. Those advantages still matter. But AI-mediated discovery creates another gatekeeper, one that may reward clear product information, sharp positioning, and relevance to a buying occasion rather than raw corporate heft. No executive should confuse that with a temporary marketing fad.
Small Referral Numbers Can Hide Large Influence
Bain says nearly 33% of U.S. consumers and 36% of European consumers use AI assistants somewhere in their shopping journey. A year earlier, those figures were 19% and 8%, respectively. That is a dramatic jump, though it deserves a level head: AI referral traffic remains below 1% of retailers’ total web traffic.
Still, dismissing it because the referral figure is small would be lazy analysis. People can ask an assistant to compare products, narrow options, summarize reviews, or identify retailers, then navigate directly to a site later. The last click may never show the assistant’s role. Bain also reports that AI referral traffic converts 90% better than non-AI traffic. That is the part that makes this worth watching. High-intent traffic, even from a tiny base, can become meaningful faster than the headline share suggests.
Market share built the old storefront; clarity may determine who gets invited into the next shopping conversation.
The immediate threat is not that agents suddenly complete every transaction. They plainly are not there yet. The threat is that a retailer becomes invisible during the recommendation phase, then wonders why traffic, awareness, and margin pressure keep getting worse. By the time a chain sees the damage in a quarterly report, the customer’s shortlist may already belong to somebody else.
Trust Still Favors the Retailer’s Own Agent
There is one encouraging finding for retailers willing to do the work. Bain says 55% of global AI-using shoppers do not fully trust AI to complete a purchase. Consumers trust retailer-built assistants at 21%, nearly twice the 12% level for third-party assistants. The gap has narrowed as third-party agent use grows, but it remains real.
That trust edge is valuable because it protects the customer relationship and the data that comes with it. Yet only 27 of the 100 largest U.S. retailers have a first-party assistant, according to Bain. Frankly, that is a remarkably low number for a group that has spent years talking about personalization like it was sacred scripture.
The Unsexy Work Is the Actual Strategy
There is no magic wand in Bain’s checklist, which is exactly why it is credible. Retailers need prompt audits around the products and occasions they care about, then they need to fix the gaps in how their value proposition appears in AI answers. Product and consumer data must be usable for personalization, not stranded in some corporate basement under a pile of incompatible systems.
- Test how third-party assistants describe the business and its priority categories.
- Make key pages speak in the language customers use for real shopping occasions.
- Set a credible first-party assistant plan, including faster white-label or outsourced options where appropriate.
- Decide who owns the coming agent-to-agent strategy before it becomes an emergency.
The holiday rush is not the time to pretend this is solved, but it is absolutely the time to find out whether a retailer is absent from the answers shoppers are already getting. Bain’s core warning is blunt enough: visibility in AI changes constantly and requires recurring attention, much like pricing or delivery. Retailers that treat it as a one-off website project are probably volunteering to be somebody else’s comparison result.