AI shopping isn't looking at your product page. It's looking at your feed

Let's skip the warm-up and get straight to it.

While half the market is busy polishing product pages — titles, bullet points, reviews, schema markup — the real battle for visibility in AI shopping is happening somewhere else entirely. In the Google Merchant Center feed. Research from Search Engine Land shows: 83% of the products ChatGPT shows users match the top 40 results in Google Shopping. And they're not pulled from the product page — they come from the feed.

That changes the whole balance of power. It reminds me of classic SEO ten years ago, when everyone stared at the text on the page and ignored the technical foundation — and a site with no proper indexing went nowhere, no matter how golden the content was. The same thing is happening now with e-commerce: your PDP (product detail page) is your storefront, but AI models read the warehouse first, not the storefront.

Why the feed matters more than you think

ChatGPT, Gemini, Perplexity don't visit your site and parse the HTML of your product page to figure out what you sell. They rely on structured data — brand, price, image, availability, category — that you explicitly submitted to Merchant Center. If that data is empty, inaccurate, or stale, the model simply never learns about that product. Doesn't matter how much copy you wrote on the page.

That's the mechanics of it. AI search doesn't read intent between the lines. It takes what's structured, clean, and current. The feed is your one direct line to the model — bypassing all the design polish and copywriting.

But don't confuse priority with replacement

This is where you need to avoid the extreme. 88% of offers in AI shopping are still pulled from product pages — meaning the PDP hasn't gone anywhere, and dropping it won't protect you, it'll weaken your position. The feed and the product page are two separate channels working together, not against each other. One gives the machine raw data for a recommendation. The other delivers conversion and trust once a real buyer lands on your site.

Whoever picks "either-or" right now loses on both fronts. Whoever holds both fronts gets visibility where the algorithm decides, and where the human decides.

What to do right now

First — go through every product in Merchant Center and make sure the data is complete: brand, decent-quality image, current price, stock status. Sounds basic, but that's exactly where visibility gets lost — a product just drops out of the pool because one field was missing.

Second — keep the feed alive. Updating it once a quarter doesn't work anymore when the decision is made by a model looking at a snapshot right here, right now. Feed freshness isn't a one-time setup — it's a process.

Third — don't overrate category structure on its own. It helps the model understand context, but it doesn't guarantee a recommendation. It's a hygiene factor, not a growth lever.

And last — get in the habit of literally checking how your products show up in ChatGPT, Gemini, and Perplexity for your key queries. That's the new visibility KPI, and almost nobody tracks it systematically yet — which means there's an easy edge here just for paying attention to detail.

Isn't it about time you started treating your feed as a real SEO asset instead of some technical formality for the IT department? I'm sure it is. And if you're ready to get systematic about this, we can talk through how to build AI search visibility for your specific business — through my services.