ChatGPT is aiming more of its searches at named domains

What happened?

Promptwatch, which tracks AI answers at scale, measured the share of ChatGPT search queries scoped to a single domain jumping from 0.37 percent to 16.8 percent in one day on August 8, 2026, while the average number of searches per answer nearly doubled from about 1.08 to 1.83. Developer Simon Willison flagged the same shift on August 20. Consultant Suganthan Mohanadasan then captured the underlying tool calls and reported that OpenAI replaced the older JSON search format with a compact one carrying a query, a freshness window in days, and an optional domain slot. OpenAI has not documented the change.

Why does it matter?

The domain slot is the part worth your attention. When ChatGPT scopes a search to a named site, it has already decided which organizations are worth checking before it searches at all. Being crawlable is no longer enough, because a query pointed at four other domains never touches yours. More searches per answer means more slots to fill, but they get filled with names the model already holds. Reddit shows how quickly this can move: Promptwatch measured the Reddit share of ChatGPT citations falling from about 3.8 percent to 0.5 percent between August 7 and August 17.

Should I do anything?

Ask ChatGPT the question a prospective member, donor, or client would ask in your category, without naming your organization. Write down which organizations it names back. That list is the shortlist your domain has to join, and the check takes five minutes and is worth repeating monthly. If you are absent, the fix is not a page tweak, it is becoming findable by name: start with one clear About page that states plainly what you do, who you serve, and where you operate.

Learn more

• Suganthan Mohanadasan on the new search format
• Promptwatch: site operator fanout data
• Our guide: how ChatGPT finds information
• Our guide: check your organization in ChatGPT