Prompts & topics
The actual questions buyers ask AI about your category: how to collect them, which ones to track, and how to turn citation gaps into content briefs — with free templates.
Prompts and topics are the raw material of GEO. A prompt is one buyer question asked to an AI assistant; a topic is the cluster of prompts around one buying intent. Collect the real questions, track which answers name you, and every gap becomes a content brief with a measurable payoff. No vendor publishes downloadable prompt tooling — the templates below are the first free set.
Where prompts come from
- Your sales calls and support tickets. Real buyer language beats any keyword tool. Mine the last 20 calls for questions starting with what, which, how, and best.
- Autocomplete and People-Also-Ask. Still useful as a seed list for the question shapes buyers use.
- Competitor citations. Ask AI comparison questions in your category and note which prompts surface rivals — those are prompts you must contest.
- Your own glossary and docs. Every term buyers must understand before purchasing is a prompt waiting to be tracked.
From gaps to briefs
Run your prompt set, mark every answer where a competitor is cited and you are not, then open the cited source. It will show you the missing piece: a comparison table, a pricing page, a definition, a review. Write exactly that page — one gap, one page, one quotable answer — and re-measure in four weeks. Challenger brands gain most here: lesser-known pages gained up to +115% visibility from targeted edits in peer-reviewed setups.
Do it in ChatBakers
- Paste the library above as tracked prompts, or generate them from your Search Console queries in one click.
- Organize into topics mirroring your funnel stages — share of voice then reads per stage, not as one blurry number.
- Let gaps surface themselves: Top Prompts and Top Cited Domains highlight exactly where rivals are cited and you aren't.
Free templates
- Prompt library spreadsheet (CSV) — 30 starter prompts across funnel stages; replace the bracketed parts with your category.
- GEO audit checklist (Markdown) — the monthly routine: re-query, gaps, briefs, re-measure.
Guides in this pillar
- 30 prompts to test your brand's AI visibility — the starter set, explained.
- From citation gaps to content briefs — the operating loop.
Related reading
Frequently asked questions
What kinds of prompts should I track?
Three kinds: category questions ('what is X'), comparison questions ('X vs Y', 'best tool for Z'), and problem questions ('how do I solve P'). Ten to twenty of each is a solid starting set of 30–60 prompts; programs in production target 20–30 prompts per core topic, re-queried monthly.
How do citation gaps become content?
When a prompt set shows competitors cited where you are not, read the cited sources: they reveal the exact subtopic, format or proof your content lacks. Each gap maps to one brief: one page, one question, one quotable answer.
Should prompts mention my brand by name?
Mostly no. Branded prompts ('what does Acme do') only test recall. The money is in unbranded prompts ('best invoice tool for freelancers') — that is where consideration sets form and where you are either named or invisible.
Check how visible your brand is in AI answers
See where ChatGPT, Perplexity and Gemini cite you — and where competitors win. Start your free 7-day trial.
Try ChatBakers free