We ran a visibility check across the seven channels that now decide whether a product gets found. Here is what came back for a typical mid-size store:
| Platform | Visibility % |
|---|---|
| Google Search | 22% |
| Google Shopping | 12% |
| Bing Shopping | 8% |
| Google SGE / AI | 45% |
| ChatGPT Shopping | 15% |
| Perplexity Shopping | 15% |
| Gemini Shopping | 15% |
Look at the spread. Google's AI surface pulls the store into answers nearly half the time. The three conversational shopping channels each land at 15%. Google Shopping itself, the channel most merchants pour budget into, sits at 12%.
That is not a rounding error. That is one channel doing three times the work of the others, and three channels where 85% of relevant shopping conversations happen without your products in them.
The 45% is not a compliment
It is tempting to read the SGE number as a win. It mostly is not.
Google's AI surfaces feed off the same Merchant Center data and product schema your store has already been maintaining for years, plus Google's own crawl of your site. You are visible there because you did the compliance work, not because you optimized for AI. It happened to you.
Which also means the 45% is your ceiling under the current setup, not your floor. And it is fragile. Google has started piloting AI performance reporting inside Merchant Center, which tells you something about where they expect merchants to be paying attention.
Why the other three sit at 15%
Each of the conversational channels sources product data differently, and most stores have set up for exactly none of them.
ChatGPT Shopping. This is the one people misread most. ChatGPT does not simply crawl your product pages and decide. Research published in March 2026, analysing 43,000 products from ChatGPT's shopping carousel, found that 83% of its recommendations matched Google Shopping's top 40 organic listings, with 60% coming from the top 10. Your Merchant Center feed is already the primary input into ChatGPT's answers, whether you knew it or not.
On top of that, OpenAI runs a merchant program that accepts a pushed product feed — CSV, TSV, XML or JSON, refreshable as often as every 15 minutes. Not crawled. Pushed. Merchants who submit accurate structured data are working with a different information layer than merchants who are only being read off Google's index.
So a 15% score here usually means one of two things: your Merchant Center feed is thin enough that you are not in Google Shopping's top 40 for your own category terms, or you have not touched the OpenAI feed at all. Often both.
Perplexity Shopping. Free merchant program, accepts Google Shopping format feeds, and Shopify stores get catalog syndication. It takes a few minutes to apply. Most merchants have simply not done it, which is why the number is low, and it is also why the number is easy to move. Perplexity skews toward research-heavy purchases, so if your products need comparison before someone buys, this channel is worth more than its user count suggests.
Gemini Shopping. Runs on Google's Shopping Graph, so in theory you should already be there. In practice Gemini rewards a different thing than blue-link Google does: it wants attributes it can reason with and compare against alternatives. Materials, dimensions, compatibility, use case, care instructions, warranty. Feeds built to pass Google's disapproval checks contain the minimum required fields and nothing else. That is enough to be listed and not enough to be recommended.
The pattern underneath all three
Every one of these systems is doing the same thing: reading structured product data, corroborating it against third-party sources, and assembling an answer. They differ in where they get the data and how often they refresh it. They do not differ much in what makes a product win.
Most feeds were built as an operational task — get products approved, avoid disapprovals, keep the ads running. Nobody built them to be the context an AI agent reasons over while comparing you against four competitors on behalf of a shopper who will never see your homepage.
The fields that matter for AI recommendation are the ones most merchants leave empty:
average_ratingandproduct_review_count— corroboration signals, and the single most common gap we findreturn_policyand shipping detail — AI answers surface these unprompted because shoppers ask about them- GTIN and brand — without them you are an unmatched product, not a comparable one
- Attribute depth in the description — the specific, boring facts a comparison needs
- Availability accuracy — stale stock data actively damages eligibility rather than being neutral
Then there is the off-site half. These models corroborate what your feed says against reviews, editorial mentions, and buying guides elsewhere. A perfect feed for a brand nobody writes about still loses to a decent feed for a brand three review sites mention.
The channels are not equally worth your time
Given the numbers above, the honest priority order for most stores:
- Fix the Merchant Center feed first. It is the single input that moves Google Search, Google Shopping, SGE, Gemini and roughly 83% of ChatGPT at once. Nothing else has that leverage.
- Apply to the Perplexity merchant program. Costs nothing, takes minutes, and the competition has not bothered.
- Submit an OpenAI product feed. Higher effort, but this is the channel with real query volume behind it — "chatgpt shopping" alone draws around 1,000 US searches a month and spiked past 2,400 last autumn.
- Check your robots.txt. If OAI-SearchBot or PerplexityBot are blocked, everything above is wasted. This happens more than you would think, usually from a scraping-protection decision made two years ago by someone who has since left.
- Then worry about content. Buying guides and comparison content give the models something to cite. But citation without a clean feed just sends the shopper to a competitor's product card.
Timeframes, roughly
ChatGPT results tend to shift within one to four weeks of feed and schema fixes. Google's AI surfaces are slower — four to twelve weeks is the realistic window. Perplexity is the fastest, because for most merchants the change is going from absent to present rather than from ranked to better ranked.
Anyone promising faster than that on Google's side is guessing.
Find out where you actually stand
The table at the top of this post is one store's numbers. Yours will be different, and the difference is the whole point — a store at 45% on SGE and 15% everywhere else has a very different problem from a store at 15% across the board.
An AI visibility audit runs your category's real shopping prompts across all seven channels, shows where you appear, where a competitor appears instead of you, and which specific feed and schema gaps are causing it.