Skip to content
Roboto
Log in Try Roboto

ChatGPT Shopping: Why Product Feeds Now Matter More Than Classic SEO

Product ranking inside ChatGPT Shopping has changed in nature. According to an observational analysis by the vendor Profound, reported by Search Engine Journal, the share of product recommendations coming from feed-integrated sources jumped from 8.26% to 61.54% on July 10. In other words, a store's visibility in ChatGPT now depends heavily on the quality and availability of its product feed, and much less on classic search ranking signals.

What the numbers say (and what they don't)

Scope matters here. The figures come from a single vendor, Profound, and cover tracked query executions, not all of ChatGPT's shopping traffic. The July analysis covers 1,757,723 query executions, with a daily sample of 50% of the month's queries, each query counted once per day. A separate 10% sample (97,725 queries between July 1 and August 24) indicates that by August, feed retrieval exceeded web search. As of September 3, Profound estimates feeds account for roughly 65% of tracked product recommendations.

These data do not show how products rank once ChatGPT pulls from feeds. Nor do they reveal how often OpenAI relies on feeds versus the open web: OpenAI does not publish that figure. This is an observational analysis, not a measurement of real buyer traffic.

A sharp shift for merchants

The effect on tracked stores is stark. Of 687 tracked clients, 450 saw their Shopping visibility drop by at least a third between July 7-9 and July 10-12. Sixty-seven clients instead recorded a gain of at least a third over the same period. Concentration is increasing: the share of references to the top 10 merchants rose from 22.5% to 41.8%, while the number of unique merchants cited fell from 13,524 to 10,607.

Among 517 clients that varied by at least 33%, a statistical model explained 83% of the variation through web search losses and feed gains. The report presents this as an association internal to its own dataset, not as proven causation.

Is GPT-5.6 to blame? A hypothesis, not a confirmation

The report attributes most of the July changes to the release of GPT-5.6 on July 9, with a rollout announced over 24 hours. OpenAI has not confirmed this link, and its ChatGPT release notes mention no shopping update on July 9 or 10. Note also that the September 3 rollout of GPT-6 Astra means the model in use today may differ from the one behind the July results. Caution is warranted about the exact technical explanation.

What this means in practice for an e-commerce merchant

The logic is simple: if ChatGPT leans more on structured feeds, then a missing, incomplete or poorly synced catalog becomes invisible, regardless of any SEO work done elsewhere. OpenAI's Help Center states that ChatGPT selects products through structured data from suppliers and merchants, without advertising or partnership influence. Feeds run through the Agentic Commerce Protocol, extended to product discovery on March 24.

Shopify and Etsy catalogs are already connected. Other merchants can request direct access, but OpenAI indicates a waitlist. The report also estimates that roughly 35% of July's feed retrieval was linked to Shopify, based on correlation of daily series.

Practical priorities

  • Check feed availability. If your platform isn't natively connected, direct access goes through a waitlist: plan ahead.
  • Invest in structured data. Titles, attributes, availability, price: these fields drive selection, not your meta tags.
  • Don't abandon SEO. The report itself ties variation to web search losses and feed gains. Both logics coexist.
  • Track ChatGPT visibility separately. Classic SEO tools don't capture this channel.

Comparison: classic SEO vs product feeds

  • Classic SEO channel: depends on content, links and domain authority; results measurable through search engines.
  • Product feeds: depend on catalog connection and structured data quality; visibility measured by third-party tools like Profound, with the limits described above.
  • Overlap: both require accurate, up-to-date product data, but a feed is a technical prerequisite, not a content signal.

Key takeaways

The trend is documented by a single, observational source: the share of feed-based recommendations surged in July, and concentration among large merchants increased. For an e-commerce merchant, the practical consequence is clear: a product feed is no longer a technical detail, it's an entry condition for ChatGPT Shopping. That doesn't make classic SEO useless, but it no longer guarantees visibility on this channel. And since OpenAI confirms neither the role of GPT-5.6 nor the frequency of feed usage, these figures are best treated as a strong signal to watch, not a settled rule.

This article is published by Roboto, a platform for generating texts, images, videos and voiceovers with AI. Roboto performed no testing on this topic.

Source

ChatGPT Shopping Results Lean Hard On Product Feeds via @sejournal, @MattGSouthern

Read next