Prompt volume

What is prompt volume?

Prompt volume is the claimed count of how often people ask a particular question in an AI assistant. Product pages use it to rank which prompts deserve tracking slots, and SEO teams meet it when keyword tools add an "AI volume" column next to a phrase.

The assistants themselves do not publish those counts. ChatGPT and Perplexity expose no API that returns "this question was asked N times today in the United States." Anything on a dashboard is built downstream: keyword indexes, consumer panels, clickstream samples, or a vendor's own prompt library, then mapped to natural-language questions.

That does not make the estimate useless. It makes it an estimate. Use it to decide which questions to track first; do not treat it as proof that an engine preferred your page on that question.

How is prompt volume different from measured AI visibility rates?

Measured rates come from reading answers to questions you fixed in advance:

• Share of voice counts answers that name your brand.
• Citation rate counts answers that link your domain.
• Each rate arrives with n, the number of answers behind it.

Prompt volume sits upstream of all of that. It is a guess about demand for a wording, not an observation of what an engine said when that wording was asked. A question with a high estimated volume can still produce zero answers naming you if your pages are unreachable or poorly structured; a low-volume question can be the one your buyers actually ask in sales calls.

Keep the categories separate in reports. "High prompt volume" belongs in a prioritization section; "cited in 4 of 12 answers" belongs in a results section.

How do vendors build a prompt volume number?

Methods vary, and vendors rarely expose enough detail to reproduce a figure. Common inputs include:

1. Keyword search volume, scaled by a factor the vendor believes maps searches to chat questions.
2. Panel or survey data about assistant use, extrapolated to a population.
3. Clickstream or referral proxies, when a vendor can see assistant traffic patterns.
4. Internal prompt libraries, when a tool seeds suggestions from its customers' tracked lists.

Each path introduces smoothing, seasonality and geography choices. Two tools showing "1,200 prompts/month" for the same string are often measuring different things. Ask what was observed, what was modeled, and how often the model refreshes.

What is prompt volume not?

It is not crawler traffic. Logs tell you a bot fetched a URL; they do not tell you how many people asked a question.

It is not AI impressions. Multiplying an estimated volume by a measured citation rate produces a third number with stacked uncertainty. Search Console may fold AI surfaces into broader search reporting; it still does not publish per-prompt chat counts.

It is not search intent on its own. Search intent is what the asker wants done; volume only suggests how many askers might exist under a wording you chose.

Why does Attensira not sell prompt volume?

Attensira's methodology is explicit: the product asks the prompts you configure on a schedule and records the answers. It does not observe what real users typed, or how many of them typed it. Selling a volume figure would imply an instrument the product does not have.

That is a feature boundary, not a missing roadmap item. Teams that need prioritization can use keyword research or a vendor that documents its volume model, then promote the winners into tracked prompts inside Attensira and read observed rates from there.

How should you use prompt volume honestly?

• Label it in slide decks and requirements docs as an estimate.
• Track a small set of buyer questions regardless of volume score, because sales and support language often differs from keyword tools.
• Promote prompts into daily reads once they matter commercially; switch from volume to citation rate and share of voice as soon as answers exist.
• Compare vendors' volumes only qualitatively unless they show inputs and dates.

When a competitor product leads with prompt volume, ask what would change in your workflow if the number were off by 2×. If the answer is "nothing," deprioritize the metric.

How does prompt volume relate to prompt tracking?

Prompt tracking in Attensira is the schedule of questions read on ChatGPT and Perplexity, not a count of human asks. The prompts are the measuring instrument: change the wording and you change what the rate measures, the same way prompt engineering changes model output.

Volume estimates can suggest which instruments to add first; they cannot replace reading those instruments once they run.

What should you read next?

• GEO measurement guide for the full measured-vs-estimated list.
• AI search performance for how observed rates behave over time.
• Reference rate for why mentions and citations are reported separately.

Prompt volume is a planning hint. The numbers that decide whether you are winning a question still come from answers, with the sample size attached.