A GEO audit is a measured answer to one question: when buyers ask AI assistants about your category, do the machines bring you up? It inventories the questions that matter for your business, samples how assistants actually answer them, records who gets named and cited instead of you, and ranks what to fix first. GEO stands for generative engine optimization, the practice of earning presence in AI-generated answers the way SEO earns presence in search results.
That is the whole definition. The rest of this page is what a real one contains, how the measurement works honestly, and what separates an audit from a screenshot.
Why this exists
A growing share of buying research now ends inside an answer instead of on a results page. ChatGPT, Claude, Perplexity, and Google's AI Overviews name a handful of brands per question, and brands outside the answer do not get a click they can lose. Classic SEO tools tell you where you rank. They do not tell you whether a machine recommends you, and most companies have never measured it once.
What a GEO audit contains
- The question set. The 10 to 25 questions your actual buyers ask assistants, in their words. Not keywords: questions. These are verified against real demand, not brainstormed.
- The sampled answers. Each question asked across the engines your buyers use, more than once. AI answers are non-deterministic: a single run is an anecdote, not a measurement.
- The mention ledger. Who gets named, per question, per engine, with counts. Your absence is recorded as precisely as your presence.
- The citation map. Which pages the assistants cite when they answer. These pages are the ground the answers stand on, and they are where positioning is won.
- The branded check. What the machines say when someone asks about you by name. Wrong or empty answers here lose deals you already earned.
- The ranked fixes. What to publish, correct, or claim first, ordered by evidence, each with the question it is meant to move.
How the measurement stays honest
Ask any vendor one question: how many samples per answer, and where is the error stated? AI answers change from day to day. Roughly a third to two-fifths of cited sources turn over daily, so any audit built on one run per question is astrology with a dashboard. A real audit states its sample sizes, its dates, and its engines on the page, and treats a thin sample as a thin sample. Ours do, and the raw answers are kept so every count can be re-derived. How that discipline works underneath, engine by engine, is the AI visibility audit page.
What a GEO audit is not
- Not a rank tracker. There are no stable "positions" inside generated answers to track.
- Not a dashboard subscription. It is a diagnosis with a date on it, and it ends in decisions.
- Not a promise. Nobody can guarantee placement inside a model's answer, and anyone who does is selling you the weather.
What it costs and what you receive
A Khanna's GEO audit is $490, delivered within a week: the question set, the sampled ledger across engines, the citation map, the branded check, and the ranked fixes, in a document written for a founder to read in one sitting. If we continue working together afterward, the audit fee is credited in full to your first month. How $490 sits against the wider market, every number sourced, is in what answer engine optimization services cost.
One honest caveat: an audit measures the starting line. Moving the answers takes publishing and positioning work measured over weeks. The audit tells you exactly where that work should start and how you will know it worked.
Common questions
How is this different from an SEO audit? An SEO audit inspects your site's health for a crawler. A GEO audit measures your presence inside AI answers, which depends on what the machines can find, quote, and trust about you, on your site and off it.
Which engines does it cover? The ones your buyers use. Typically ChatGPT, Claude, Perplexity, and Google AI Overviews, stated explicitly in the report with per-engine results, never averaged together.
We are small. Is it worth measuring? Small is the best time. Answer real estate in most B2B categories is still lightly contested, and the first brand to earn a citation for a question tends to keep it while competitors wonder why their traffic curve went flat.
Can we run it ourselves? Partly. Ask your top ten buyer questions in two engines and note who is named: that is a useful afternoon. The audit's value is the sampling discipline, the citation map, and knowing which absence to fix first.
To get one, write to daksh@khannasllc.com with your domain in the subject line. That is the entire intake process.