Updated on 19 November 2026, by João Leite, Founder of murmur.

ChatGPT search volume exists only as a tool estimate, and each tool estimates it in its own way. Semrush, which sells an SEO tool, provides AI Volume, the estimated number of searches each topic gets in an AI model, and Profound, which sells AI monitoring, builds its number from licensed conversations with AI models. Profound’s conversations come from ChatGPT, Gemini and Perplexity and go through cleaning and probability modeling, according to the Prompt Volumes documentation. Anyone who came wanting to know how many searches ChatGPT gets in total, against Google, will find that calculation in the guide on whether AI models are replacing Google.

  • Question volume in AI models exists as a tool estimate. Each tool arrives at the number through its own method, and the tool itself describes the method.
  • Semrush estimates by topic. AI Volume is the estimated number of searches for a topic in a chosen AI model, alongside the estimated monthly activity of related topics.
  • Profound starts from licensed conversations. Prompt Volumes uses conversations with ChatGPT, Gemini and Perplexity, cleans the data, applies probability modeling and shows the region of each question.
  • Ahrefs weights by regular search. In Brand Radar, the visibility of each question is weighted by the regular search volume behind it.
  • AI volume and Google volume do not move together, in Profound’s calculation. On 16 December 2024, Profound wrote that AI and Google Search patterns have little correlation.

ChatGPT search volume is a tool estimate

ChatGPT search volume is an estimate that the tool calculates, not a count that the buyer verifies. On Google, search volume is the number of times a word is searched per month, and the SEO market has used that number for years. In AI models, the person writes the question in their own words, in a conversation, and monitoring tools arrive at volume through different paths.

Three tools show the path they take. Semrush estimates volume by topic, Profound starts from licensed conversations with AI models, and Ahrefs uses regular search volume to weight the visibility of each question. Each number is built on a different dataset, and that is why one tool’s number cannot be checked against another’s.

Semrush: AI Volume, estimated by topic and by month

Semrush shows search volume in ChatGPT and the other AI models as AI Volume, a number estimated by topic. Semrush’s help page for the prompt research report defines AI Volume as “The estimated number of search queries the topic gets on a selected AI Platform”, the estimated number of searches the topic gets in the AI model the person chooses. The same page shows the volume of related topics, which is the estimated monthly AI search activity around them.

Semrush’s unit is the topic, not the exact question. The number speaks to the whole topic, not to one way of writing the question. The question database Semrush relies on, and Ahrefs’ warning for brands with little search, are in the guide on how Semrush and Ahrefs measure AI citation. How Semrush uses real prompts to measure the brand, alongside Peec AI, is in the guide on Peec AI or Semrush AI Visibility.

Profound: volume from licensed conversations with AI models, with the region of each question

Profound builds question volume from real conversations with AI models, which it licenses. In the documentation, Prompt Volumes is described as a dataset made from licensed conversations with answer engines, with the data cleaned and probability modeling applied on top. The sources of the conversations are ChatGPT, Gemini and Perplexity.

Region is part of Profound’s data. The same documentation says the tool serves to identify which regions drive each search, and that makes it possible to see where interest in a question comes from. For anyone selling in Brazil, it is worth checking which region the number comes from before using it.

Ahrefs: visibility weighted by regular search volume

Ahrefs weights brand visibility in AI models by the regular search volume behind each question. On the Brand Radar page, Ahrefs’ tool for tracking brands in AI models, the estimated impressions line says “potential visibility weighted by the real search volume behind each prompt”. Ahrefs sells an SEO tool.

The volume that weighs on Ahrefs’ calculation is regular search volume, not conversation with the AI. A question with a lot of regular search weighs more heavily in the brand’s estimated impressions, and one that almost nobody types into a search engine weighs little.

The three ChatGPT search volume estimates side by side

The three ChatGPT search volume estimates start from different bases, and each tool describes its own on its own page, read on 9 November 2026.

what is comparedSemrushProfoundAhrefs
who it issells an SEO toolsells AI monitoringsells an SEO tool
where the number comes fromestimate of searches by topic, in a chosen AI modellicensed conversations with ChatGPT, Gemini and Perplexitythe real regular search volume behind each question
the unitthe topic, with the monthly activity of related topicsthe question, with the region that drives itthe brand’s potential visibility, in estimated impressions
what the tool says about the methodestimated numbercleaned data and probability modelingvisibility weighted by search volume

All three give estimates: two estimate question volume in AI models, each with its own base, and the third uses regular search volume as a weight. The price of each tool is in the guide on how much AI visibility monitoring costs.

In Profound’s December 2024 calculation, the question in the AI and the search on Google have little correlation

In Profound’s calculation, published on 16 December 2024, question volume in AI and search volume on Google barely move together. In the launch of Prompt Volumes, Profound wrote that “AI and Google Search patterns show little correlation.” It is the reading of those who sell the question volume data, made almost two years ago.

Profound’s calculation and Ahrefs’ method pull in different directions. Ahrefs uses regular search volume to weight the questions asked to AI models, and Profound wrote that the two patterns look little alike. For anyone building the question list, the practical consequence is not to treat Google volume as AI volume without looking at where each number comes from.

How to use volume in the question list without discarding the niche question

Estimated volume serves to see which topics the buyer asks AI models about most, and the question list is not cut by it. A niche question may have no number in any tool and still be the one the buyer asks before hiring. Search volume does not decide how many pages to publish, as the guide on how many pages to publish for your company to appear in ChatGPT shows, and where to get the questions from is in the guide on how to build the question list to measure the brand.

In practice, volume goes in as a column of the list, alongside the question, the category and who asks it. An empty column does not remove the question from the list. It only signals that no tool has an estimate for that way of asking.

At murmur, the list comes from mapping the sector’s questions: the FotoRegistro case study

A murmur client’s question list comes from mapping the sector’s questions, not only from a volume estimate. murmur, a GEO agency in Brazil that places brands in the answers of ChatGPT, Claude, Google AIO and other AI agents, publishes the methodology in four steps, in a cycle that does not stop: mapping the sector’s questions, the diagnostic, the production and structuring of content and ongoing monitoring. The mapping comes first, and the contract sets no limit on it: the work continues with no cap on questions, hours or content.

The FotoRegistro case study, a photo printing and personalization e-commerce, shows the mapping moving along with the market. On 2 September 2026, 84% of the new questions that appeared in the market already had a FotoRegistro article answering them.

Frequently asked questions

Can you know how many people asked ChatGPT an exact question?

What the tools show is an estimate, not a count of the people who asked an exact question. Semrush estimates volume by topic, grouping the questions about the same subject, and Profound reaches its number with probability modeling over licensed conversations with ChatGPT, Gemini and Perplexity. In both cases, the number is the tool’s reading, with the method it describes.

Does question volume on ChatGPT follow Google volume?

In Profound’s account, published on 16 December 2024, no: the patterns of AI and of Google search had little correlation. That is the view of the one selling the question volume data, and from almost two years ago. Ahrefs, on the other hand, weights each question’s visibility by regular search volume, so each number is read with the method of the tool that calculated it.

Is a question with no volume worth a page?

It can be worth it: search volume does not decide how many pages to publish, and the genuinely different question the buyer asks calls for a page with the complete answer, whether or not there is a volume number next to it. The criterion is in the guide on how many pages to publish for your company to appear in ChatGPT.

Does question volume change from one country to another?

Profound shows the difference by region: the Prompt Volumes documentation says the tool identifies which regions drive each search. So, when reading a question volume, check which region it is from before using it for the Brazilian market.

Where to start

Before looking at volume, see what AI models answer today to the questions in your market. The free diagnostic from murmur takes your site URL and shows, without sign-up and in under two minutes, how AI models answer people looking for what you sell, with the full report by email. murmur’s sector studies show which brands AI models recommended in the sectors it measured.

Read also

References