According to Resoneo, since switching to GPT-5.3 Instant in early March, ChatGPT responses pointed to about 20% fewer sites.
The analysis comes from the French SEO consultancy and draws on data from Meteoria, an AI visibility tracking platform that monitored 400 prompts daily for 14 weeks and returned 27,000 comparable responses.
The average unique domains per response fell from 19 before the transition to 15 after. The average unique URLs per response fell from 24 to 19.
The URL-per-domain ratio remained at 1 throughout the tracking period. The data suggests that ChatGPT doesn’t visit as many sites per response, but examines each one just as deeply.
Now fewer domains share the same citation surface in each answer, meaning the sites that are cited take up a larger share of each answer.
Server logs secure the pattern
An independent protocol analysis by Jérôme Salomon at Oncrawl supports the results. Its data tracks ChatGPT users’ bot activity across multiple websites and shows that crawl volumes have plateaued at lower levels. Some pages are no longer crawled at all and the crawl frequency for pages that are still visited has decreased.
Resoneo attributes the change to the fact that ChatGPT’s default experience is now more driven by GPT-5.3 Instant, which the company says triggers fewer web searches and citations than previous behavior. Oncrawl’s server log data shows the lower crawling pattern over the same period.
Previous data has shown that AI platforms cite sources differently than traditional search. An SE ranking analysis of 129,000 domains found that referring domains were the strongest predictor of the likelihood of a ChatGPT citation, with a threshold effect at 32,000 referring domains.
A Search Atlas report also showed little overlap between Google rankings and ChatGPT citations, with a median domain overlap of around 10-15%.
Why this is important
A 20% decrease in cited domains per answer means fewer sites competing for visibility in each ChatGPT answer. The total citation space shrank, but the sites that continued to appear maintained their crawl depth.
For anyone tracking ChatGPT referral traffic, the early March model transition is a date range to check in your analytics.
Looking ahead
Resoneo’s analysis notes that GPT-5.4 Thinking reintroduces search fanouts and uses site: operators to target trusted domains. However, these behaviors were not captured in the quantitative dataset covering GPT-5.3 Instant and below.
It is not yet clear whether the citation area will continue to narrow or expand again with newer models.
Follow us on Facebook | Twitter | YouTube
WPAP (907)
