Key points
- Analysis by Konitzny of 1,000 cases found ChatGPT gets partner content as ready-made summaries, not live page fetches.
- Other websites reach the model as raw text snippets, and partner entries sit earlier in the prompt than any web result.
- In ranked lists, summaries mention 70 percent of the top three items but just 23 percent of items ranked seventh to tenth.
Why it matters for marketers: If partner content gets a structural head start inside ChatGPT, brands outside those deals face a tougher fight for visibility, so teams should track how and where they are cited.
A new analysis by Konitzny, reported by The Decoder, describes an uneven setup behind ChatGPT's answers. Content from OpenAI's partners is not retrieved live. The model is handed prepared summaries of those pages, and these are positioned ahead of every ordinary web result in the prompt.
Pages from the rest of the web get less polished treatment. They are passed to the model as short unprocessed text excerpts. So before ChatGPT writes a word, partner material has already been condensed and placed first, while other sources arrive in rougher form and later.
The summaries also seem to skew toward the top of ranked lists. Across 1,000 cases Konitzny examined, they named 70 percent of the items in the first three spots, but only 23 percent of the items in spots seven through ten. The source does not say why.
The report adds context. OpenAI promised partners more prominent placement in ChatGPT in 2024, and a separate study from Hamburg found Springer media showing up much more often in the web interface than through the API. The Decoder also notes that even partners are not always cited correctly.
For marketers and content teams, the takeaway is that visibility in ChatGPT is not purely a contest of content quality. Publishers without a partnership may be competing from a weaker position. It is worth checking how your brand is cited, comparing it with partner outlets in your category, and watching whether OpenAI responds to these findings.
Sources
This brief is our own summary. Read the full reporting at the source.
