OpenAIChatGPT AI Search Updates — March 2026

On March 5-6, ChatGPT stopped exposing query fanouts in its responses — the "Searching N queries" indicator disappeared. This chart shows 60 days of fanout data leading up to that change, capturing the moment ChatGPT went dark on its web search behavior. Based on data tracked by Promptwatch.

Average Fanouts Per Response — Jan 5 to Mar 6

The average number of web search queries ChatGPT generated per response from Jan 5 to Mar 6. On March 5–6, fanouts dropped to zero — ChatGPT stopped exposing its query fanouts in responses.

What this means for you

Two changes are visible in this chart. First, a slow squeeze: average fanouts slid from 2.03 queries per response in early January to 1.84 by March 4 — ChatGPT gradually consulting the web less per answer. Then, on March 5-6, the fanout data vanished entirely as ChatGPT stopped exposing its search queries in responses. A core piece of AI search transparency — the exact queries used to retrieve and cite sources — was removed overnight, without any announcement.

How to act on it

• Export and archive the AI search metrics your reporting depends on. When a platform stops exposing a signal, the historical baseline you saved is the only one you'll ever have.
• Don't read the drop to zero as ChatGPT abandoning multi-query search — it stopped showing its queries, not necessarily running them. Cross-check behavior shifts against independent signals like crawler logs and citation counts before changing strategy.
• Keep covering topic sub-questions. Right up to the cutoff, ChatGPT was still averaging over 1.8 searches per prompt, so content that addresses multiple angles of a topic remained multiply retrievable.

How we collect this data

We collect millions of prompt responses, citations, and click data from the actual user interfaces of major AI platforms: over 26 billion data points and growing. This gives us one of the largest datasets on how AI search engines cite sources and recommend brands.

Real UI monitoring

Data straight from the interfaces of ChatGPT, Gemini, Perplexity, Claude, AI Overviews, and more.

26B+ data points

Over 26 billion analyzed citations, prompts, and responses, one of the largest AI search datasets available.

Continuously updated

Refreshed constantly so the trends you see reflect the latest behavior of AI search engines.

Aggregated & public

Published freely for the GEO community, based on aggregated, non-identifiable trends.