AI shopping behavior barely moves between categories. Electronics tops the table, but it beats home and garden by under four percentage points — the opposite of what most merchants assume.
Two research firms measured AI shopping behavior this autumn and came back with numbers that look incompatible. NIQ reports that 51% of US consumers used an AI-powered shopping tool in the past month. Meanwhile, Similarweb puts AI in 11.4% of shopper journeys.
So which is right? Both are. They count different things, and the gap between them is the most useful part of either study.
But the finding that should change what you do this week is smaller and quieter. It sits in the category breakdown.
Plug In SEO checks whether your product titles and descriptions actually carry the terms you are targeting.
What the AI shopping behavior data actually says
Similarweb and Statista broke their journey figure down by retail category in The State of Ecommerce 2026. Here is the whole spread.
| Category | Share of journeys involving AI |
|---|---|
| Consumer Electronics | 14.1% |
| Beauty | 13.2% |
| Fashion | 11.7% |
| Ecommerce overall | 11.4% |
| Marketplaces | 11.0% |
| Home & Garden | 10.4% |
Top to bottom, that is a range of 3.7 percentage points. The category AI touches least still sees it in roughly one journey in ten.
Scope matters here, so take it plainly. These are Similarweb clickstream figures for June 2026, US desktop only. A journey counts as AI-involved when an AI tool appears somewhere in the path to a transactional retail site. The overall 11.4% was 4.5% in July 2024, so the direction is steep even though the category spread is flat.
51% of US consumers report using at least one AI-powered tool to support their shopping in the past month — the first time adoption has crossed the halfway mark. Product recommendations lead at 20% adoption, personal shopping assistants at 16%.
NIQ, Agentic Commerce Tracker, 24 September 2026. Survey of approximately 500 US consumers. Source
That NIQ figure deserves its caveat stated out loud. Roughly 500 respondents is a small sample, and it is self-reported rather than measured. Treat it as a read on attitudes, not a traffic forecast.
Why the two numbers disagree
NIQ counted people. Similarweb, by contrast, counted journeys. As a result, a shopper can answer yes to “did you use an AI shopping tool last month” after a single query, then buy eleven other things without touching AI once.
So the honest translation is this. About half of your customers have tried it, and roughly one purchase path in nine currently runs through it. Neither number says AI has replaced anything.
That also explains why the category spread stays narrow. AI is being used early and broadly, for ideas and shortlists, rather than deeply inside one vertical. Because it is a habit rather than a specialism, it spreads evenly.
What AI shopping behavior means for your store
1. Write for the comparison, not for the AI
Electronics and beauty lead this table for the same reason. Both are bought by comparing attributes — wattage, capacity, shade, ingredients, finish. So the stores that lose here are the ones whose product copy describes a mood instead of a spec. Our guide to SEO for electronics stores goes deeper on why spec-heavy titles still miss how buyers actually search.
2. Put the comparable facts in text
After all, a spec trapped in a product image cannot be read, compared or quoted. So move sizes, materials, compatibility and measurements into the page body or into structured product fields. Then the same facts serve a shopper skimming on mobile and a parser building a shortlist.
3. Keep price and availability consistent
In practice, shoppers arriving from a recommendation arrive late. We covered that timing effect in the seven-day tail an AI recommendation leaves behind. If your price or stock has drifted since the shortlist was built, the visit converts into a bounce.
4. Treat AI as every category’s problem, not a side project
Because the spread is flat, there is no category where this is a niche experiment and none where it can be safely ignored. So the useful question is not which categories justify attention. Instead, it is whether your product information is accurate and structured enough to be quoted correctly wherever it surfaces. That groundwork is what every AI surface draws on, and it deserves real resourcing rather than being treated as an experiment.
What not to conclude from this
Do not read 10.4% as permission to ignore it. Home and garden sits at the bottom of this table, yet AI still appears in one journey in ten there, and the overall figure has more than doubled since July 2024. A floor is not an exemption.
Equally, do not read 14.1% as a reason to rebuild your store around assistants. The same report found that 89% of AI-involved journeys also use Search. So the page that wins an AI shortlist is, in almost every case, the page already working in Google.
And do not quote NIQ’s 51% as though half of all orders now come through AI. Instead, it measures whether someone tried a tool, not where their money went.
Where to go next
Selling in one of the two categories at the top of the table? Our beauty-store guide covers shade-variant duplication and claims language, which are the two things that most often make a beauty catalogue hard to compare. For the underlying report, Similarweb publishes The State of Ecommerce 2026 in full.
Frequently asked questions
Does AI shopping behavior really differ by product category?
Barely. Similarweb’s June 2026 clickstream data puts consumer electronics highest at 14.1% of journeys and home and garden lowest at 10.4%. That is a spread of 3.7 percentage points across the whole retail landscape, so no category is meaningfully exempt and none is dominated.
If 51% of consumers use AI, why does it appear in only 11.4% of journeys?
The two studies count different units. NIQ asked about 500 US consumers whether they had used any AI shopping tool in the past month, which half of them had. Similarweb measured individual purchase journeys. One person can use AI once and then complete many unrelated journeys without it.
I sell home and garden. Should I do anything differently?
No, and that is the point of the finding. Your category sits 1.0 percentage point below the ecommerce average, not outside the trend. The same work — comparable facts in text, accurate price and stock, valid product structured data — applies at 10.4% as it does at 14.1%.
Do I need a separate AI search strategy?
Not a separate one. Similarweb found that 89% of journeys involving AI also involve Search, so the two overlap almost entirely. Treat it as one job: a product page that answers a comparison clearly tends to serve both surfaces at once.
Plug In SEO scans your product and collection pages and flags where titles, descriptions and structured data are missing the terms you are trying to rank for.