OpenAI says search inside ChatGPT now runs at about a fifth of Google's query volume. The definition is narrow, the Google baseline undisclosed, and referral traffic tells a different story. I take the number seriously anyway: if a fifth of questions are answered inside a chat window, the writer's job stops being ranking and becomes being the citable artefact. My June 2026 note on Answer Engine Optimization, llms.txt and crawler access already says what to do. Concrete ask: publish the crawlability audit of my ~6,800 posts, and treat 1.2 billion weekly ChatGPT users as an audience reachable only by being quotable and dated.
OpenAI's consumer product lead for ChatGPT, Adam Fry, told a Monday evening dinner in New York that search volume inside ChatGPT is now "roughly one-fifth (20%) of the volume of queries of Google." The Verge reported it on 3 November 2026; Search Engine Journal traced it the same day to the transcript Alex Kantrowitz of Big Technology published afterwards.
I want to take that seriously, and then ask what it counts. Fry's rule is narrow by design: "Every time you ask a question and go search the web, that's a search." That excludes every prompt answered from memory. And he did not say which Google figure he compared against, over what period, or in which regions (SEJ, 3 November 2026).
The other measure points the other way. Across 118,671 sites tracked in Ahrefs Web Analytics in October, Google accounted for 24.74% of traffic and ChatGPT for 0.36%, according to the tracker's 3 November 2026 update. Query volume and referral traffic describe two different worlds, and quoting one without naming which is which is quietly selling something.
Why does the query number still matter to a writer? Because a fifth of questions answered inside a chat window is a fifth answered where there is no result page, no rank, no click. The contest stops being about position and becomes about selection: whether the engine treats your post as the artefact worth naming.
That is where I have been heading for years. In August 2025 I reprinted my own 2010 line — that no one would in future "search for 'information' on Google" but would simply state a problem and receive a solution — and added my 2022 guess that Search Engines would morph into Solution Engines (24 August 2025). In December I read OpenAI's search plans as echoes of my past (18 December 2025).
What I had not done, until my correspondence of June 2026, was treat it as a plumbing problem. There the discipline was named plainly — Answer Engine Optimization, being chosen as the definitive answer rather than ranking for a keyword — and the advice was unglamorous: publish llms.txt, put a verifiable date behind every claim, and let the crawlers in. If robots.txt shuts out GPTBot, ClaudeBot, PerplexityBot and Google-Extended, nothing else you do matters. So, concretely, this month:
- A crawlability audit of my ~6,800 posts: which are indexed, what canonical each one claims, and which are blocked without my knowing.
- An llms.txt seeded with the themes I actually own, and dated structured data so that the date — my only prior-art evidence — becomes machine-checkable.
- A weekly scoreboard: ask each engine the questions I claim to have answered first, and publish whether I was named, including the weeks I was not.
1.2 billion people used ChatGPT in a week, according to Big Technology's report as summarised by SEJ. To most of them I am invisible, and being quotable and dated is the only door I have. Citation is never guaranteed. An open door, and a way to measure the attempt, is all I am asking for.
With regards,
Hemen Parekh
www.Iam-Immortal.ai / www.IndiaAGI.ai / www.YourContentCreator.in
4 Nov 2026
No comments:
Post a Comment