AI Mode is a conversational search mode where the answer stands on its own and there are no organic results next to it. For an online store that changes the shape of the opportunity: it isn’t about your position in a list, it’s about being one of the sources the answer is built from. What decides is the data a customer actually cares about — price, availability, specs.
That sounds like a big change, and in one respect it is. In another it’s boring: the rules stay the same, it just shows more plainly when the data is sloppy.
What changes and what doesn’t
| Aspekt | Regular results | AI Mode |
|---|---|---|
| What the customer sees | A list of stores, they pick themselves | An assembled answer and links to sources |
| What's at stake | Position in the list | Being one of the answer's sources |
| Comparison query | The customer opens several tabs | The system assembles the comparison from several sources |
| What decides | Relevance and position | Available, consistent product data |
The conditions for showing up don’t change. Google’s own documentation says there are no special requirements for AI features and no separate markup — the same foundation applies as for regular search. Among its recommendations it also notes that important content should be in text form and that structured data should match the visible text. That’s covered in more detail in how to appear in AI Mode.
Shopping queries are comparison queries
Here’s what matters for a store. Queries like “which vacuum is best for pet hair” or “what should I use on my paint after winter” aren’t about one product — they’re about criteria. The answer is assembled from several sources, and each one contributes whatever it has that the others don’t.
A store carrying nothing but the manufacturer’s description has nothing to contribute. Your competitors have the same copy.
How to write a product page that clears that bar is covered in product pages for AI; why a readable price is a particularly sensitive spot, in pricing on your site and AI.
How much traffic this actually brings — our data
So this isn’t purely theoretical: on our own store we measured what AI tools really send. Over twelve months it was 1,867 visits, 90 percent of them from ChatGPT. Their conversion rate was 3.5 percent against 0.88 percent from Google organic, and close to three times the revenue per visit.
So: small volume, higher-quality visit. The full data and the methodology are in the AI traffic case study.
Why it’s hard to measure
This is the uncomfortable but honest part. AI Mode can’t be isolated from the data that’s commonly available:
- The Search Console report is kept for generative AI features in aggregate, not per surface.
- Assistant traffic gets mixed together in analytics, and in our case ChatGPT accounted for ninety percent of it.
- Citations also shift between individual generations — earlier research cited in an Ahrefs study puts it at roughly 45 percent — so a one-off check proves nothing.
In practice that means the claim “we get X visits from AI Mode” would be hard to back up today — and so would any offer that promises that result.
Common mistakes
Treating AI Mode and AI Overviews as one thing
They’re different surfaces with different citations — in an Ahrefs analysis from September 2025 (US), the cited URLs overlapped by 13.7 percent. A number from one tells you nothing about the other.
Passing data from one surface off as the other
It happens most often with Search Console impressions, which are reported in aggregate.
Relying on the manufacturer's description
It adds nothing to a comparison that your competitors don’t also have. The value is in what you add.
A mismatch between the page and the feed
When price or availability contradict each other, the system has nothing to work from and looks elsewhere.
Where this article stops
It doesn’t give market-specific AI Mode numbers. None are publicly documented and availability is rolling out gradually, so we’d rather leave them out than estimate.
Our data is an illustration, not a benchmark. One store, one period, the state before optimization.
It doesn’t promise sales. It describes what changes and what you can influence. Anyone promising specific revenue from AI Mode is promising more than anyone can back up today.
It doesn’t cover shopping agents in general. For those, see AI shopping agents and online stores.
What to take away
For an online store, AI Mode doesn’t change the rules — it changes the shape of the opportunity. Instead of a position in a list, it’s about being one of the sources of an assembled answer — and only someone with something to compare gets into the comparison.
The volume is still small. Our data shows 1,867 visits over twelve months at a single store, but with a much better conversion rate than organic. That’s a good reason to pay attention and a bad reason to bend your whole budget around it.
And above all: be careful with numbers. Even we can’t separate AI Mode performance from the rest today — and anyone claiming they can should show the methodology first. For the wider context, see SEO for e-commerce in the AI era.
Want to know whether your product data has what an assembled answer needs? Sniper Design runs AI SEO audits that go through product pages, structured data, and page-to-feed consistency, and help name the next steps.
Data sources: our own store’s measurement over twelve months (published July 17, 2026) and Google Search Central documentation on AI features.
