Skip to content

Issue 32 ·

What decides placement in ChatGPT's organic and paid results

What decides placement in ChatGPT's organic and paid results

I ran a query in ChatGPT last week - "running shoes for flat feet".

An ad from FIGS appeared below the response.

FIGS makes medical scrubs. The product in the ad was the FIGS x New Balance 997H collab, a sneaker built for healthcare workers who spend long shifts on their feet.

The organic result above it pulled three shoes: the ASICS GEL-KAYANO, the Brooks Adrenaline GTS, and the HOKA Gaviota. All three are stability shoes built for overpronators, which is what flat-footed runners need.

ChatGPT found the right answer above the line. Below it, FIGS ran an ad for a shoe with nothing to do with running. The ad copy said "comfort, performance, style" and named none of what a runner with flat feet is looking for.

ChatGPT Ads don't work on keywords

OpenAI runs a relevance-weighted, second-price auction where relevance multiplies bid - a higher CPC alone does not move a slot. The system scores relevance on four things: the context of the conversation, the landing page, the ad title, and the ad copy. The most important of those is what the landing page carries as structured catalog attributes, because that is what the system reads before the ad enters the auction at all.

For FIGS to have matched on a flat feet query, the landing page would need attributes that describe who the 997H is built for - healthcare workers, people standing on hard floors for long shifts - not just how the shoe looks. The product is right for that person. The catalog does not say so in a form the system can read.

ChatGPT does not use different systems to populate the two slots. Both are resolved against the same structured data in the Shopify Catalog feed.

ASICS, Brooks, and HOKA results surfaced organically because their catalog entries carry attributes that signal stability and overpronation support.

FIGS won the auction, but the catalog data behind the 997H is tuned to a different buyer and a different context, so it surfaced for the wrong query.

If you fixed your catalog for ChatGPT ads eligibility, those same fixes have already moved your organic match rate. The fields that get you through the 100-product sample review are the same fields that determine how relevant your placement is for ads.

And if your product page copy says the right things but the structured field underneath is empty, neither slot finds you - because the catalog is not in a form the system reads.

- Ankit

If this was useful, the next one will be too.

Weekly. Free.

Unsubscribe with one click.