AI shopping is here. Even Gymshark only captures 0.1% of it.

In January 2026, Adobe published the most consequential ecommerce data point of the year, and almost nobody outside the analytics community blinked.

That number, measured across more than one trillion visits to U.S. retail sites, is the cleanest signal we have that AI shopping has crossed the threshold from "interesting experiment" to "channel that matters." On Cyber Monday alone, AI traffic to retail jumped 670% YoY. November ran +769%. December +673%. Every other industry trailed: travel +539%, financial services +266%, tech +120%, media +92%.

Retail moved fastest because retail is where the question "what should I buy?" gets asked.

And it's not just more traffic. It's better traffic.

By Q1 2026, the trend accelerated again. Adobe's April 2026 quarterly report shows AI traffic up another 393% YoY, with conversion now running 42% better than non-AI sources. That's a complete reversal from a year earlier, when AI traffic converted worse than paid search.

Salesforce's data tells the same story from a different angle: brands on Salesforce Commerce Cloud saw a 119% YoY increase in traffic from AI assistants in Q4 2025. Gartner now forecasts that 22% of all global Cyber Week orders will be driven by intelligent agents. Bain projects 15-25% of total ecommerce flowing through agentic channels by 2030. McKinsey puts the number at $5 trillion globally.

If you read the macro reports, the conclusion is obvious: AI shopping is the next acquisition channel, and the brands that get there first will compound a structural advantage.

So a fair question is: how much of this are real brands actually capturing today?

To answer that, let's look at one of the best-run direct-to-consumer brands in the world.

Gymshark: a control group for "doing it right"

Gymshark is a £1.45B fitness apparel brand built almost entirely on direct-to-consumer ecommerce. As of 2025, 96% of its revenue came through its own web stores and apps. It's run by Ben Francis, who returned as CEO in 2021 specifically to sharpen the brand's data and analytics edge.

Gymshark is exactly the kind of brand that should be capturing AI shopping traffic. They have:

• A modern Shopify-based commerce stack
• A documented partnership with Google Cloud, where Ben Francis has publicly described using AI to "absorb customer sentiment instantly to shape product design," "power smarter pricing and go-to-market strategies," and "scale globally with a strong data foundation"
• An enterprise HawkSearch deployment for AI-powered site search, which their own published case study credits with a 23% lift in average order value, 20% conversion lift, 66% more search sessions, and 52% more new users

In other words, Gymshark is not the laggard. Gymshark is the brand that other brands benchmark against.

So if you want to know what "doing AI commerce well" looks like in 2026, you measure Gymshark.

Here's what you find.

Read that twice. One of the most sophisticated DTC operators on earth, with a heavy AI investment thesis, captures roughly one in every thousand visits from the AI surfaces that everyone else's deck calls "the next acquisition channel."

This is not a Gymshark problem. It's a catalog problem.

The visibility gap, in two numbers

The market is doing one thing. Brands are doing another. The two are not yet connected:

The gap isn't the consumer. The consumer is already there. The gap is the merchant infrastructure that connects a buyer's question to a brand's catalog.

MetaRouter's 2026 agentic commerce report calls this "the visibility gap" and is brutally direct about the cause:

"Consumer demand for AI shopping is real (39% adoption, 805% traffic growth), but conversion lags because merchant infrastructure was not built for agents. The 4.4x conversion potential exists, but only for merchants whose data infrastructure can support agent decision-making."

Why even Gymshark can't muscle through this

You might assume that if you spend enough on AI infrastructure, this fixes itself. Gymshark is the proof that it doesn't.

Their AI investment is inward-facing:

• HawkSearch improves site search, what happens after a buyer is already on gymshark.com.
• Google Cloud AI improves internal sentiment analysis, pricing, and product design.

Both of those are excellent. Both of them lift the conversion rate of the ~10.83M visitors who somehow already arrived. Neither of them affects whether ChatGPT, Perplexity, or Google AI Overviews can confidently cite a Gymshark hoodie when a user asks "what's a good lifting hoodie for cold gyms?"

That citation is decided upstream of the brand's website entirely. It's decided by:

1. Whether the AI can resolve "Gymshark Crest Hoodie" to a clear product type
2. Whether structured attributes are present (material weight in GSM, fit, fabric composition, gender, intended use)
3. Whether variants are enumerated with size, color, price, availability
4. Whether the data is consistent with what's syndicated through Google Merchant Center, Shopify's product graph, and any future Salesforce-style ChatGPT integrations
5. Whether the AI sees enough provenance to trust citing the brand

If any of those break down, the AI defers to a competitor whose data is cleaner, even if the competitor's product is worse. You can have the better hoodie and lose the recommendation.

What the math looks like in 2027

Take Adobe's growth rate at face value. Generative AI traffic to retail grew 693% in 2025. Even if it cools dramatically, say to a 3x annual run rate as the base gets bigger, here's what happens to a Gymshark-sized brand if catalog readiness doesn't improve:

That's the base case. If you also believe Adobe's 31-42% conversion-rate premium (and there's no reason not to), by 2028 the AI channel is contributing more high-intent revenue than several of Gymshark's existing acquisition channels. By 2029 it's a top-three source.

But that math only holds if the AI can find and cite the products. If your catalog is unreadable, the curve doesn't compound. You stay at 0.1% while a more readable competitor curves up to 5%.

What you actually have to do

This is the un-sexy answer. There is no AI button to press. There is only catalog work. Specifically:

1. Resolve every product to a clean type. "Hoodie" is not a type. "Men's heavyweight pullover hoodie, brushed cotton, gym/training" is.
2. Capture foundation attributes as structured fields. Not in marketing copy. In real attribute: value pairs that survive being passed through a ChatGPT plugin call or a Salesforce Commerce Cloud catalog feed.
3. Enumerate variants completely. Every size × color × in-stock × price combination, with GTINs where available.
4. Add provenance. When did this fact get verified? Where did it come from? AI systems are increasingly weighting evidence by source. "Last verified by merchant on 2026-03-12 from supplier datasheet" is worth far more than "extracted from product description paragraph."
5. Treat the catalog as the AI surface, not the website. Because that's what AI shopping reads.

This is what Listwiser does. We extract, resolve, and verify foundation attributes across your catalog, surface the blockers, and apply one-click fixes with full audit trails, so by the time AI shopping is 5% of your traffic instead of 0.1%, your products are the ones it can confidently cite.

The window

The Salesforce / OpenAI ChatGPT integration entered pilot in April 2026. Gartner predicts 22% of Cyber Week orders will be agent-driven this year. Adobe is recording quarterly growth rates that compound to 5-10x channel share within 24 months.

The brands that fix their catalog this year will compound. The brands that wait will look up in 2027 and find the AI-cited shelf full of competitors, and the only honest reason will be that the competitors' data was readable first.

Gymshark, for what it's worth, has the runway, the team, and the infrastructure to fix this faster than almost anyone. The question is whether they (and you) start now.

---