Prompt volume
The AI-era analogue of search volume — an estimate of how often a question gets asked of the assistants. Useful, but directional.
Prompt volume is an estimate of how often a particular question or prompt is asked of AI assistants — the AI-era analogue of keyword search volume. It is meant to help you prioritise: a query asked often is worth being cited on. The honest caveat is that prompt volume is directional, not gospel. Unlike search engines, the AI vendors do not publish per-prompt query counts, so every prompt-volume figure on the market is a modelled estimate built from proxies (search data, sampling, panels), not a measured number. Treat it as a relative signal — this query is probably bigger than that one — rather than a precise traffic forecast. The metric you can verify directly is not how often a query is asked, but whether you are actually cited on it, which is what reverse AI search measures.
What is prompt volume meant to tell you?
In classic SEO, search volume helps you decide which keywords are worth chasing. Prompt volume tries to do the same job for AI search: if buyers are constantly asking a model “what’s the best X for Y?”, then being named in that answer is valuable. As a prioritisation aid it is reasonable.
Why is prompt volume only directional?
Because the underlying data does not exist publicly. The model providers don’t release how many times each distinct prompt is asked, so vendors estimate it from proxies. That means:
- The number is modelled, not measured — accuracy varies by method.
- Prompts are infinitely varied — people phrase the same intent dozens of ways, which is hard to bucket cleanly.
- It changes fast — AI usage patterns shift quickly as behaviour and models evolve.
We pressure-test the metric with our own index in does prompt volume mean anything? Our stance: use it to rank priorities, not to forecast traffic.
What should you measure instead?
Measure the thing you can verify: whether you are cited, on which queries, by which models. That is your AI citation footprint and your AI share of voice — both grounded in real model answers rather than estimated demand. Run the free Domain Check to see the real queries a domain is cited on today, then use prompt-volume estimates only to decide which of the gaps to close first.
Worked example
Take two prompts a travel brand might want to win: “best time to visit Japan” and “best time to visit a small town few people have heard of.” Even without exact numbers, the first is clearly higher prompt volume — it is backed by far more related search demand — so earning a citation there is worth more. Notice, too, that re-phrasing matters: “when should I go to Japan” and “Japan travel season” are the same intent but would be estimated separately, which is exactly why prompt volume is a directional ranking aid, not a precise count.
Related terms
- AI query index — the data layer that records which queries actually cite you.
- Reverse AI search — the verifiable counterpart to an estimated prompt volume.
- AI share of voice — how you weight prompt-volume estimates against competition.
Frequently asked questions
Is prompt volume the same as search volume?
It is the AI-era analogue. Search volume counts how often a keyword is searched; prompt volume estimates how often a prompt is asked of an AI assistant. They are related but not identical, and prompt volume is much harder to observe directly.
Can you measure prompt volume exactly?
No — AI providers do not publish per-prompt counts. Every figure on the market is a modelled estimate built from proxies, so prompt volume is best treated as a relative signal rather than an exact number.