ChatGPT Search Now Uses the site Operator at Scale
Promptwatch data shows ChatGPT Search adopting the site: operator overnight on August 8, 2026, with searches per response nearly doubling at the same time.
On August 8, 2026, ChatGPT Search changed how it searches the web overnight. Fanout queries using the site: operator, searches scoped to a specific domain, jumped from 0.37 percent to 16.8 percent of all fanout queries in a single day. At the same time the average number of searches per response nearly doubled. ChatGPT now runs more, and more targeted, background searches per answer.
The site operator fanouts report, published August 10, 2026, tracks the shift day by day. For anyone who cares about showing up in ChatGPT answers, the practical question is what a domain scoped search actually means for your chances of being cited.
The measured change
Promptwatch's report covers two series. The first is the percentage of all ChatGPT Search fanout queries that contain the site: operator, per day. The share hovered between 0.3 percent and 0.5 percent for weeks. It dipped briefly to 0.15 percent on August 3 to 5, which is consistent with a staged rollout or a pre launch experiment. Then it jumped to 16 to 17 percent on August 8. That is roughly a 46x increase in share within a single day.
The second series is the average number of fanout queries per ChatGPT Search response, per day. On August 8 the average jumped from about 1.08 to about 1.83, the same day the site: usage surged. The timing is the key signal. The new domain scoped searches come on top of the generic web searches ChatGPT was already running, they do not replace them. If ChatGPT had swapped generic queries for site: queries, the average would have stayed flat.
What the data does and does not tell you
The chart shows that the shift happened in a single day. It does not say why. A jump that fast, in lockstep across the fanout behavior, points to a model or system prompt rollout on OpenAI's side rather than a gradual behavior change. Platform controlled search behavior can change overnight, which is the part that makes one off audits unreliable and continuous monitoring necessary.
What the report cannot tell you is which domains ChatGPT targets with the site: operator. The data is aggregated and non identifiable. It shows the share of queries that are domain scoped, not which domains those queries target. For your own brand, the question is whether ChatGPT is running site:yourdomain.com against your topics, and that is a question only your own tracking answers.
Why this matters for visibility work
The shift changes what it takes to be cited. Before August 8, a site: scoped search was a rounding error, about 1 in every 280 fanout queries. Since August 8, roughly 1 in 6 fanout queries is scoped to a specific domain. ChatGPT is no longer just searching the open web and seeing what comes back. It is deliberately going to specific sites to pull information from them.
That has two effects. The first is that your domain is now a retrieval surface in its own right. When ChatGPT runs site:yourdomain.com [topic], it is searching your site directly. Thin category pages, broken internal search, and unindexed content now cost you answers on top of rankings. The second is that the average answer now draws on more distinct retrievals, so there are more chances to be found, but each chance is more targeted.
What to do about it
The first move is to treat your own domain as a retrieval surface. The pages ChatGPT can reach with a site: query are the pages it has indexed from your domain. Make sure your most important pages are crawlable and indexed. A site: scoped query can only surface what search engines have indexed from your domain, so gaps in coverage translate directly into gaps in AI answers.
The second move is to fix the parts of your site that a site: query exposes. Internal search is the obvious one. If your on site search is broken, a site:yourdomain.com query returns poor results, and that shows up in the answer. Category pages that are thin or duplicate are the same problem. The pages that rank well in Google are not always the pages that serve a domain scoped AI query well.
The third move is to re run your fanout analysis for your tracked prompts. The queries ChatGPT generates from a prompt changed materially on August 8. Keyword mappings built before the shift are stale. A fanout you mapped in July no longer matches the fanout ChatGPT runs today, so the prompt to keyword map needs a refresh.
How to measure it properly
A population report like Promptwatch's tells you what ChatGPT is doing across the whole web. It does not tell you what ChatGPT is doing for your prompts. For that you need your own fanout data joined to your own visibility data.
That join is the part most visibility tools skip. A prompt tracker that reports whether ChatGPT mentions your brand will not tell you whether ChatGPT runs a site: query against your domain, how many fanout queries your prompts trigger, or how the fanout depth changed after August 8. The fanout volume and depth is the measurement that turns a population report from interesting context into something you can act on.
Promptwatch publishes that data as query fanout tracking. The feature tracks query fanout volume, average queries per response, and query length trends over time, based on ChatGPT Search data. For a report that shows the site: shift at the population level, the matching move is to open the same fanout view for your own prompts and see whether the shift shows up in your fanout depth, and whether the queries ChatGPT runs for you changed on August 8.
The broader pattern
The site: shift is one instance of a wider change in how AI search retrieves information. The average searches per response nearly doubling at the same time is the other half of the story. ChatGPT now searches both wider, more queries per answer, and deeper, queries scoped to specific domains, than it did before.
For visibility work, that means the retrieval surface is getting more granular. A year ago, the work was to be present in the open web results ChatGPT returned. Now the work is to be present in the open web results and in the domain scoped results, and the domain scoped results are a new, additional path into the answer that did not exist before.
The practical response is to treat fanout depth as a first class signal. The brands that notice when their prompts start triggering more queries, when those queries start targeting specific domains, and map the new fanout to the citations that follow, are the ones that stay visible through a shift like this. The ones that only watch the visibility score will see the site: shift as a change in ChatGPT's behavior, not as a change in their own visibility.
What to watch next
The report is a snapshot through August 17, 2026. The open questions are whether the site: share holds, whether the average searches per response keeps climbing, and whether the new domain scoped queries translate into different citation patterns. Those are exactly the questions a fanout view answers for your own prompts.
The dataset Promptwatch publishes is aggregated and non identifiable, and it is refreshed constantly. That makes it useful for spotting population level shifts like this one. It does not replace the need to track your own brand, your own prompts, and your own fanout depth. The two work together: the public report tells you what is happening in the field, and your own tracking tells you whether it is happening to you.
For a team that wants to act on the site: shift rather than just read about it, the workflow is to open the public report for context, open the fanout view for your own prompts, and connect the two. That is the measurement stack that turns a fanout shift into a visibility decision, and it is the stack Promptwatch is built around.