What is prompt volume?
Prompt volume is an estimate of how many times people ask AI assistants about a given topic in a period, usually a month. It plays the role in AI search that keyword search volume plays in SEO: it tells you which questions are worth trying to be the answer to.
For example, a tool might report that prompts about "choosing a CRM for a small business" run at a certain monthly level across ChatGPT and other assistants. That figure would cover many differently worded prompts, from "what CRM should a 5-person agency use" to "is HubSpot overkill for a small team", grouped into one topic. That grouping is the first big difference from search volume.
Every prompt volume figure on the market today is an estimate. No AI platform publishes how often a given prompt is asked.
Where does prompt volume data come from?
Prompt volume data comes from third-party models, not from the AI platforms. As of October 2026, OpenAI, Google, Anthropic and Perplexity do not offer a public tool, like a keyword planner, that reports how many times a prompt or topic is asked. Tools that sell prompt volume build it from their own inputs, typically some mix of panel or clickstream data, search volume for related keywords, and modeling, and they differ in what they disclose about the method.
The closest first-party data is indirect:
- Bing Webmaster Tools. Bing's AI Performance report, launched in public preview on February 10, 2026, shows citations and "grounding queries", the searches an AI system ran before citing a site [1].
- Google Search Console. Since June 16, 2025, data from AI Mode counts toward the totals in the Search Console Performance report [3], blended with ordinary search data. Google's separate generative AI report, announced in June 2026, shows impressions by page, country, device and date, with no clicks and no query-level breakdown [1] [2] [7].
Neither of these is prompt volume. Bing shows the searches an AI ran, and Google shows how often a site appeared, but neither shows how many people asked a given question.
What is the difference between prompt volume and search volume?
Search volume counts how often an exact keyword is typed into a search engine, while prompt volume estimates how often a topic is asked about in AI assistants, across countless wordings. The two behave differently enough that one cannot be converted into the other.
| Search volume | Prompt volume | |
|---|---|---|
| Unit | Exact keyword | Topic or intent cluster |
| Typical length | A few words | A sentence or a paragraph, often with context |
| Source | Search engine data, such as ad planner tools | Third-party estimates |
| Published by the platform? | Yes, in rounded ranges | No |
| Follow-ups counted? | Each search counted separately | Varies by tool, often unclear |
| Can be checked against your own data? | Partly, via Search Console | Barely |
The unit is the core problem. A keyword like "crm for small business" is typed the same way by thousands of people. Prompts are long and personal, so almost every exact wording is unique. A count for one exact prompt would be close to zero and meaningless, which is why prompt volume tools group prompts into topics. How a tool draws those topic boundaries changes the number it reports.
What is the difference between prompt volume and grounding query volume?
Prompt volume counts questions people ask, while grounding query volume counts searches an AI system runs to answer them, and one prompt can trigger several searches. Gemini, for example, decides whether a prompt needs search and then "automatically generates one or multiple search queries and executes them" [4]. Google's AI Overviews and AI Mode may use "query fan-out", issuing related searches across subtopics [5] [6].
So a figure for grounding queries, like the one in Bing's report [1], measures machine searches, not human demand. A single prompt about choosing a CRM might produce separate searches for pricing, integrations and reviews. Reading grounding queries as prompt volume overstates demand for the subtopics and hides the original question.
This is the distinction most guides miss, and it is the one that matters when a vendor's numbers come partly from search data. A prompt volume estimate built on search activity may be counting the AI's searches as well as the people's.
How accurate is prompt volume?
Prompt volume has no published accuracy benchmark, because there is no first-party total to check it against. A keyword search volume figure can at least be compared with a site's Search Console impressions for a ranking keyword. A prompt volume figure for ChatGPT has no equivalent check outside the platform.
That does not make the metric useless. It makes it relative. Prompt volume is most defensible for comparing topics measured the same way by the same tool (topic A is asked about more than topic B) and least defensible as an absolute count. Treat any precise-looking prompt volume figure as a modeled estimate and ask the vendor three questions:
- What is the raw input? Panel data, clickstream data, search data, or modeling from something else.
- Which assistants does it cover? ChatGPT alone, or Gemini, Perplexity, Copilot and Google's AI features too.
- How are prompts grouped into topics? The grouping decides the number.
How did prompt volume emerge as a metric?
Prompt volume emerged as AI assistants became a place people ask commercial questions, and as AEO tools needed a way to choose which prompts to track. The idea is borrowed directly from keyword research, but the data has lagged the demand for it, with platforms releasing only partial AI reporting in 2025 and 2026.
2025: AI Mode enters Search Console totals
On June 16, 2025, Google said data from AI Mode now counts toward the totals in the Search Console Performance report [3]. AI search activity became visible to site owners, but blended with ordinary search and not separable by prompt.
February 2026: Bing reports grounding queries
Bing Webmaster Tools launched its AI Performance report in public preview on February 10, 2026, with citations, grounding queries, page-level citation activity and trends [1].
June 2026: Google adds AI impressions
Google announced generative AI performance data in Search Console alongside a new AI Search opt-out on June 2, 2026, rolling out first to a subset of UK site owners [7]. The report shows impressions but no clicks and no query-level detail [1] [2].
How do you use prompt volume?
Use prompt volume to rank topics against each other, not as a forecast of traffic. AI answers often satisfy the question without a click, so even an accurate prompt count does not translate into visits the way search volume roughly does.
A practical workflow:
- Start from your own keyword data. Search Console queries and keyword research show which questions your audience already asks. AI prompts on the same topics are usually longer versions of them.
- Group by intent, not wording. Build a topic list ("compare CRMs for agencies", "CRM pricing for small teams") and track AI answers per topic, not per exact phrase.
- Use prompt volume to set priority. Spend tracking and content effort on the topics a tool rates highest, then check whether you are cited.
- Check grounding data where you can get it. Bing's grounding queries [1] show what AI systems actually search for when they cite sites.
- Report it as an estimate. In any client or management report, label prompt volume as modeled and name the tool, so no one mistakes it for a platform-reported figure.