GEO isn't the new SEO. It's SEO for your catalog

Everyone is optimizing content for AI. Almost nobody is optimizing the catalog.

GEO is the acronym of the moment. Generative Engine Optimization: showing up in answers from ChatGPT, Gemini and Google’s AI Mode. And nearly the whole conversation happens inside content marketing. Rewrite the blog, adjust the tone, earn citations, build an FAQ page.

For anyone who sells products, that’s the easy part.

The agent doesn’t read your page

When someone asks an agent to “find me a neutral running shoe, size 9, under 100 dollars, delivered within three days”, the agent isn’t browsing your storefront. It needs to answer four questions, and answer them fast:

Does the product exist, and is it available? Exposed, reliable inventory, not the storefront’s promise.
Does the product match the requested attributes? Is “neutral” in a structured field, or buried in the middle of a description paragraph?
Is that really the price? The final price, with the promotion applied, not the list price.
Can it be delivered in time? Shipping that can be quoted by postal code without going through checkout.

If any of those answers lives only in the page’s HTML, you’re not in the answer. You’re invisible to exactly the one making the purchase decision.

Catalog debt always existed. Now it charges different interest

Almost every e-commerce operation knows its product data is poor. Attributes in the wrong field. Descriptions inherited from the supplier. Variants registered as separate products. Spec sheets published as images.

That was always debt. It cost conversion in on-site search, filters that don’t filter, returns from wrong expectations. And it never moved up the priority list, because the cost was diffuse: a bit of conversion here, a bit of customer service there.

In agent-mediated buying, the cost stops being diffuse. A product the machine can’t read doesn’t lose a little conversion. It simply doesn’t make the list of options.

It used to cost conversion. Now it costs existence.

The yardstick moves

For twenty years, the e-commerce yardstick was the on-site experience: speed, navigation, product page, checkout. It still matters for whoever reaches the site. But a growing share of the decision will happen before that, somewhere the site doesn’t appear.

The question becomes a different one: is your catalog machine-readable?

And that changes who owns the problem. Content GEO is marketing work. Catalog GEO is platform and product data work: the product data model, attribute quality, exposure of inventory and price, shipping quotes decoupled from checkout, structured data published consistently.

It’s the least glamorous team in the operation that decides whether the brand shows up in the answer.

A cheap test before any investment

Before hiring a GEO tool or agency, run a test that costs one afternoon.

Take your five best-selling products. For each one, ask: could an external system, without opening the page, answer availability, attributes, final price and delivery time?

If the answer is no for any of the five, you don’t have a content problem. You have a data problem. And no blog rewrite fixes a data problem.

What changes in practice

Product data moves up the priority list. It stops being an operational chore and becomes a distribution asset.
Attributes become a contract. Whatever is searchable has to be in a structured field, with a controlled vocabulary.
Price and inventory need to be queryable from outside. With the same truth checkout uses.
Shipping stops being checkout’s secret. Delivery time by postal code has to be quotable before purchase intent.

None of this is new. All of it was “important, but not urgent”. Agentic commerce is what turns that old backlog into urgency.

Where does product data sit on your priority list today?