Query Fan-Outs, Prompt Volumes, and Difficulty Scores in ChatGPT and Gemini

A prompt list copied from a keyword export is how GEO programs waste a quarter. Buyers do not type the same strings they type into Google. We build the list in Promptwatch from sales language, then we use volumes, difficulty, and query fan-outs to decide what stays.

Essential is $95/mo and 50 prompts. Kick-off is $199/mo with unlimited prompts when we hold several logos. Explore is free: 10 ChatGPT prompts, no Gemini. We say that in the kickoff so nobody thinks the demo covered Gemini.

Volumes and difficulty are filters, not targets

Promptwatch attaches monthly search volumes, difficulty scores, and citation rates to prompts across ChatGPT, Claude, Gemini, and Perplexity. We treat volume as a way to drop trivia, not as a Google Keyword Planner clone. Difficulty tells us the prompt is crowded. It does not tell us we should write a 4,000-word page.

Custom prompts still matter more than the explorer dump. Comparisons, "best X for Y," alternatives, and the pricing questions sales already hears. Prompt Explorer can import topics from Google Search Console and generate rows from People Also Asked. We use that to fill holes. We do not let it replace the list the account lead already knows.

Topics and tags keep a 50-row list readable. Personas sit on the same prompts so "IT buyer" and "CFO" are not mixed in one average. Paid plans refresh daily. We do not promise instant alerts.

Fan-outs are the retrieval list

Query fan-outs show how a model expands the typed prompt into background searches. Promptwatch stores the exact fan-out queries behind each response, plus terms and themes, with a per-model compare. ChatGPT and Gemini often disagree on those background searches. That is useful. The page we are missing is usually on the fan-out, not on the typed string.

If the model searched a definition or a comparison we do not have, that is a brief. See how to get cited by ChatGPT for the passage work after the brief exists. If the fan-out is a review URL or a Reddit thread, we do not write a twin article. We decide whether a mention on that source is in scope.

Otterly from $29 will not give you this join. Gemini is an add-on there, and the data can lag. Peec at $95 is a clean mention tracker. Profound Starter at $99 annual is ChatGPT only. Fan-outs are why we stay on Promptwatch for the list, not why we delete Ahrefs for Google.

Paid engines on Essential and up include ChatGPT, Gemini, Claude, Perplexity, Grok, Llama, DeepSeek, Mistral, Copilot, AI Overviews, and AI Mode. We still start the conversation with ChatGPT and Gemini because those are the two names most stakeholders ask about. The fan-out compare is where Gemini earns its seat.

What we will not do with the scores

We will not build a page for every fan-out variant, or skip a revenue prompt because difficulty is high. Volume on an assistant is not Google search volume. G2 has Promptwatch at 4.7/5. That does not make the volume number a forecast.

Professional is $245/mo when the list grows past 50 or a second project appears. Business is $579/mo for larger in-house teams. We do not quote a $29 Promptwatch plan.

FAQ

Can we import the whole Search Console query report?

You can import topics into Prompt Explorer. We still freeze 25 to 50 prompts that sound like a buyer, not like a long-tail dump. A 400-row list is unreadable in a QBR.

Do fan-outs exist on Explore?

Explore is ChatGPT only and 10 prompts. You can show the idea. You cannot run a ChatGPT-versus-Gemini fan-out program on that tier.

How we run it

Collect 25 prompts from sales calls and pricing pages. Tag them.
Check volume and difficulty. Drop the trivia. Keep the revenue questions even if difficulty is high.
Read fan-outs for the ten prompts that are absent or citing someone else.
Open one brief for an on-site gap, or one outreach note for an off-site source.
Email hello@1001seomedia.com if you want us to freeze the list. Bring the last month of "what did they ask ChatGPT" anecdotes from sales.