The prompt panel
Two hundred commercial prompts written from your buyers' language, not your product vocabulary. Documented and handed over, so you can re-run it yourself or hold us to it.
What a GEO audit measures
A GEO audit measures four things: whether an engine names you, whether it links you, which sources it pulled from, and who it recommends when it does not recommend you. Those four together give a share-of-voice figure you can track, rather than an impression of how things are going.
There is no rank tracker for generative engines. Ask ChatGPT the same question twice and the wording changes; ask it from a different account and the sources can change too. That variability is exactly why most agencies avoid measuring this and sell you activity instead of a number.
The way around it is sampling. Run enough prompts, enough times, across enough engines, and the noise averages out into something stable enough to trend. A single answer is an anecdote. Two hundred prompts run weekly is a measurement.
So the audit is not a screenshot of ChatGPT saying something nice about you. It is a panel: a fixed, written-down set of the questions your buyers actually type, executed cold with no personalisation, scored the same way every time. Fixed matters more than large — a panel you change halfway through cannot show a trend.
The uncomfortable half of the output is the competitor half. When an engine answers a buying question and does not name you, it names somebody. Knowing who, and which of their pages the model pulled from, is usually the most actionable page of the whole report.
What you receive
You receive the prompt panel itself, a scored visibility baseline per engine, a source analysis showing which pages models retrieved, a competitor share-of-voice table, and a prioritised list of the gaps worth closing first. The panel is yours to keep and re-run.
Per-engine visibility score
Mentioned, cited with a link, or absent — scored separately for ChatGPT, Perplexity, Gemini, Copilot and Google AI Overviews. They disagree more than you would expect, and the differences are instructive.
Source analysis
Which URLs the models actually retrieved when answering. Frequently these are third-party listicles and forum threads rather than anyone's product pages, which changes where you should spend.
Competitor share of voice
Who gets named when you do not, how often, and on which prompts. This is the section clients circulate internally.
Gap analysis
The prompts where you are closest to breaking through, separated from the ones you will not win this year. Effort against plausible return, ordered.
Re-run methodology
Written scoring rules so a re-run six months from now is comparable to this one. Without it you have a snapshot, not a baseline.
How the audit runs
The audit runs over twelve working days: prompt design with your sales team, a cold run across five engines, source and competitor analysis, then a live readout. Your time is roughly two hours, most of it spent telling us how buyers actually phrase things.
- 01
Prompt design
We sit with whoever talks to your buyers and collect the questions they genuinely ask. Sales calls are a better source than your keyword tool here, because generative queries are conversational and long.
→ A written 200-prompt panel, approved by you
- 02
Cold run across five engines
Every prompt executed without personalisation, logged in / logged out, across ChatGPT, Perplexity, Gemini, Copilot and AI Overviews. Cold matters — a logged-in run reflects your own history, not a buyer's.
→ Raw responses captured and scored
- 03
Source and competitor analysis
For every answer, what was cited and who was named. We cluster the sources to show which kinds of pages the models trust for your category, and which of your competitors have quietly become the default answer.
→ Source clusters and a share-of-voice table
- 04
Readout and roadmap
A live walkthrough with your team, then a prioritised gap list. If the honest answer is that your constraint is classic SEO or content depth rather than GEO, we say so and point you there instead.
→ Board-ready readout and an ordered gap list
How a GEO audit differs from an SEO audit
An SEO audit asks whether a search engine can crawl and rank your pages. A GEO audit asks whether a language model retrieves and repeats them. The inputs overlap — both need crawlable, well-structured content — but the measurement, the competitor set and the winning content shape are different.
| SEO audit | GEO audit | |
|---|---|---|
| Question answered | Can engines crawl, render and rank this site | Do models retrieve, cite and repeat this brand |
| Measurement | Rankings and impressions from Search Console | Share of voice across a fixed prompt panel |
| Unit of analysis | The page | The passage, and the third-party source that mentions you |
| Competitor set | Whoever ranks for your keywords | Whoever the model names — often not the same companies |
| Typical fix | Architecture, internal linking, page depth | Answer-shaped content, entity signals, third-party presence |
| Feedback speed | Weeks, via a rank tracker | Weekly, via a re-run of the panel |
Signals you need this now
You need a GEO baseline when AI referral traffic appears in analytics, when sales hear buyers quoting an assistant, when a competitor keeps getting named in answers, or simply when nobody in your business can say whether models mention you at all.
- Analytics shows referrals from chatgpt.com, perplexity.ai or copilot
- Buyers arrive already holding a shortlist you were not on
- Sales report prospects quoting an assistant's summary of your category
- A competitor is consistently named in AI answers and you are not
- You are about to invest in content and want a before number
- Your board has asked what your AI search strategy is
- Nobody can currently answer whether models cite you
What the baseline tends to show
Most brands start lower than they expect. Engines lean on third-party sources — comparison posts, forums, review sites — far more than on your own pages, so companies with strong classic SEO are often close to invisible in generative answers until that gap is closed.
Where this fits
Related GEO services
Common in these sectors
The adjacent disciplines
Deciding between options
Questions about GEO audit
Each answer is written to stand alone in 40 to 60 words — the shape an AI Overview or Perplexity citation lifts. Ships with FAQPage schema.
How is a GEO audit different from an SEO audit?
An SEO audit asks whether search engines can crawl and rank your pages. A GEO audit asks whether language models retrieve and repeat your brand when a buyer asks a question. Both need sound technical foundations, but the measurement and the winning content shape differ substantially.
Can you really measure something as variable as ChatGPT?
Yes, through sampling. Any single answer is unreliable, so we run a fixed panel of two hundred prompts across five engines and score the aggregate. The individual responses vary; the share-of-voice figure across the whole panel is stable enough to trend week over week.
How many prompts do you test?
Two hundred commercial prompts as standard, written from your buyers' actual language rather than your product vocabulary. Larger panels are available for broad catalogues. What matters more than size is that the panel stays fixed, otherwise later runs are not comparable.
Which engines do you cover?
ChatGPT, Perplexity, Gemini, Microsoft Copilot and Google AI Overviews as standard. They disagree with each other more than most people expect, which is itself useful — being strong in one and absent from another usually points at a specific, fixable source gap.
What happens after the audit?
You get a prioritised gap list and you are free to action it yourself; the panel and methodology are handed over. Clients who continue with us move to an ongoing programme with weekly panel re-runs, but the audit is deliberately sold as a standalone diagnostic.
Is it worth auditing if we have no AI traffic yet?
Usually yes, because the absence is the finding. Citation work is slow — it depends on third-party mentions accumulating — so a brand with zero presence today benefits most from starting to measure now, before a competitor becomes the entrenched default answer.
Find out whether this is your constraint.
Thirty minutes with a senior strategist. We pull your live visibility while we talk and tell you plainly whether a GEO audit is what you need — or whether your problem sits somewhere else.