How to Run a GEO Audit: The Complete 7-Step Methodology

A GEO audit is a structured assessment of how visible your brand is in AI search engines — whether ChatGPT, Perplexity, Gemini, and other AI platforms cite you when users ask questions about your category. Running one involves seven steps: define your query set, check technical crawler access, query each AI platform and log citations, analyze competitor citations, audit your content for extractability signals, score the results, and prioritize fixes. A thorough manual GEO audit takes 2-4 hours; an automated audit returns the same diagnosis in about 60 seconds.

You cannot improve what you have not measured. Before investing in GEO — content restructuring, schema, brand-mention building — you need a baseline: where does your brand actually stand in AI search today? A GEO audit answers that. It tells you whether you are cited, for which queries, how you compare to competitors, and what specifically is holding you back.

This guide is the complete manual methodology. Follow all seven steps and you will have a full picture of your AI search visibility — or run an automated audit to get the same diagnosis in a fraction of the time.

Last updated: May 2026

A GEO audit produces the baseline every GEO program needs: are you cited, for which queries, versus which competitors, and why or why not. The manual methodology is seven steps and 2-4 hours of work. The discipline of running one — rather than guessing — is what separates GEO programs that work from GEO effort that is wasted on the wrong fixes.

What a GEO audit measures

A GEO audit measures four things: citation presence (are you cited at all), citation prominence (where in the response), citation quality (how you are described), and competitive position (who is cited instead of you) — plus the technical and content factors that explain those results.

A complete GEO audit answers two layers of question. The outcome layer: how visible are you in AI search right now? The diagnostic layer: what is causing that result?

The outcome layer measures:

• Citation presence — Across your target queries, what percentage result in your brand being cited at all? This is the headline number.
• Citation prominence — When you are cited, where in the AI response do you appear? First-cited carries far more weight than buried in a footnote.
• Citation quality — How are you described? "[Brand] is the leading X" is a strong citation; a bare URL in a source list is weak.
• Competitive position — Who gets cited when you do not? Which competitors dominate your category in AI search?

The diagnostic layer explains the outcome by checking:

• Technical factors — crawler access, Bing indexation, page speed
• Content factors — extractability, structure, factual density, schema
• Authority factors — brand mentions, entity signals, third-party presence

The outcome layer tells you where you stand. The diagnostic layer tells you what to fix. A useful audit produces both — a score and a prioritized fix list.

A GEO audit has two layers: the outcome (citation presence, prominence, quality, competitive position) and the diagnosis (the technical, content, and authority factors that explain the outcome). An audit that produces only a score without the diagnosis tells you that you have a problem but not how to fix it.

How to run a GEO audit: the 7 steps

The seven-step manual GEO audit: (1) define a target query set, (2) check technical crawler access, (3) query each AI platform and log citations, (4) analyze competitor citations, (5) audit content for extractability, (6) score the results, (7) prioritize fixes by impact.

Step 1: Define your target query set

Identify 15-30 queries that matter to your business — the questions a potential customer would ask an AI when researching your category. Include a mix of:

• Category queries: "best [category] for [segment]"
• Comparison queries: "[you] vs [competitor]", "[competitor] alternatives"
• Problem queries: "how to [solve the problem your product solves]"
• Brand queries: "what is [your brand]", "is [your brand] good"

Write the queries the way a real person would phrase them to an AI — full, conversational sentences, not keyword fragments. This query set is the spine of the entire audit; everything else measures against it.

Step 2: Check technical crawler access

Before measuring citations, confirm AI search engines can actually access your content. Check your robots.txt for any rules blocking GPTBot, ChatGPT-User, PerplexityBot, ClaudeBot, Google-Extended, or Bingbot. Confirm your site is indexed by Bing (ChatGPT and Copilot retrieve through Bing) and ideally by Brave Search (Claude's backend). A site that fails this step has a hard ceiling on citation regardless of content quality — this is the most common single cause of zero AI visibility.

Step 3: Query each AI platform and log citations

Run every query from your set through each major AI platform — ChatGPT (with browsing), Perplexity, Google (AI Overviews and Gemini), and Claude (with web search). For each query × platform combination, log: was your brand cited, where in the response, how was it described, and which other brands appeared. This is the core data-gathering step. For 20 queries across 4 platforms, that is 80 data points — budget 60-90 minutes.

Step 4: Analyze competitor citations

From the data in Step 3, build a competitive picture. Which competitors are cited most often across your query set? Which queries do they dominate? What content of theirs is being cited — comparison pages, original research, review aggregator profiles? Competitors' cited content is a roadmap: it shows you what AI search engines reward in your specific category.

Step 5: Audit your content for extractability

For the pages that should be cited but are not, audit the content against the known citation signals: Does it lead with a direct answer? Does it use definitive, specific language with named entities and data? Is it structured into clean, self-contained sections? Does it have FAQPage and Article schema? Is it recently updated? Each missing signal is a candidate fix. See our AI citation signals guide for the complete signal list.

Step 6: Score the results

Convert the data into a score. Weight citation presence most heavily (it is the fundamental threshold), then prominence, quality, and competitive position. A 0-100 score — a GEO Score — gives you a single trackable number. Even a rough score is valuable because it makes progress measurable across re-audits.

Step 7: Prioritize fixes by impact

Finally, turn the diagnosis into an ordered action list. Fixes that unblock citation entirely (crawler access, Bing indexation) come first — they are prerequisites. Then content extractability fixes on your highest-value pages. Then authority and brand-mention work, which compounds over months. Order by impact and by how many queries each fix affects. The output of the audit is this prioritized list.

The seven-step methodology turns a vague sense of "are we visible in AI search" into a concrete baseline plus a prioritized fix list. The discipline matters: most wasted GEO effort comes from skipping the audit and optimizing the wrong things. Steps 1-4 measure; steps 5-7 diagnose and prioritize.

How often to run a GEO audit

Run a full GEO audit quarterly, with lighter monthly check-ins on your top 10 queries. AI models update, retrieval systems evolve, and competitors optimize — a GEO audit is a recurring measurement, not a one-time diagnostic.

GEO is not static. Three things change continuously: AI models retrain and update, retrieval systems evolve, and competitors optimize their own content. A citation you hold today can be lost next month to a competitor who published better-structured content. So GEO auditing is recurring, not one-time.

The recommended cadence:

• Quarterly: full audit. Run all seven steps across your complete query set. This catches drift, surfaces new competitive threats, and measures the impact of the previous quarter's fixes.
• Monthly: light check-in. Re-run your top 10 queries across the main platforms. This is a 20-30 minute spot-check that catches significant changes early.
• After any major content change. When you publish or substantially rewrite key pages, re-audit those pages' target queries 3-4 weeks later to measure the change.
• After AI platform updates. When a major AI platform announces a significant model or retrieval update, run an off-cycle audit — these updates can shift citation patterns substantially.

Track your GEO Score across every audit. The trend line — not any single score — is what tells you whether your GEO program is working.

A GEO audit is a recurring measurement: quarterly full audits, monthly light check-ins, plus off-cycle audits after major content changes or AI platform updates. The GEO Score trend across audits — not any single number — is the real measure of whether your GEO program is working.

Manual audit vs automated audit

A manual GEO audit takes 2-4 hours and gives you deep, qualitative understanding of your AI search position. An automated GEO audit returns the same core diagnosis in about 60 seconds and scales to far more queries — most teams use a manual audit once to learn the methodology, then automate ongoing measurement.

Both approaches have a place:

The manual audit

Running the seven steps by hand takes 2-4 hours for a 20-query set across four platforms. The value is depth — you see exactly how each AI describes your brand, read the competitor content being cited, and develop an intuitive feel for your category's AI search dynamics. Every GEO practitioner should run at least one manual audit to genuinely understand the methodology. But it does not scale: at 30+ queries tracked weekly, manual auditing becomes an unsustainable time sink.

The automated audit

An automated GEO audit runs the same query-and-log process programmatically across multiple platforms, scores the results, and returns the diagnosis in about 60 seconds. It scales to large query sets, runs on a schedule, and removes the variance problem (it can query each prompt multiple times and aggregate). The tradeoff is less qualitative depth — you get the data, not the felt sense of reading 80 AI responses yourself.

The practical approach

Most effective teams do both: run one manual audit to learn the methodology and build intuition, then use an automated audit for ongoing measurement at scale. The LumenGEO free audit runs the core methodology automatically — query set across ChatGPT and Perplexity, citation logging, competitor analysis, and a scored result — in about 60 seconds, no signup required. It is the fastest way to get the baseline that Step 1 through Step 6 of the manual process produce.

Run one manual GEO audit to learn the methodology and build category intuition — then automate ongoing measurement. The manual audit gives depth; the automated audit gives scale and consistency. The free LumenGEO audit produces the core diagnosis in about 60 seconds versus 2-4 hours by hand.

Frequently asked questions

What is a GEO audit?

A GEO audit is a structured assessment of how visible your brand is in AI search engines. It measures whether ChatGPT, Perplexity, Gemini, and other AI platforms cite your brand when users ask category questions, how prominently, versus which competitors — and diagnoses the technical, content, and authority factors behind those results.

How long does a GEO audit take?

A thorough manual GEO audit takes 2-4 hours for a 20-query set across four AI platforms — most of that is Step 3, querying each platform and logging citations. An automated GEO audit returns the same core diagnosis in about 60 seconds.

How many queries should a GEO audit cover?

15-30 queries for a solid audit. Fewer than 15 produces a noisy picture where a single query swings the result; more than 30 is valuable but time-consuming to do manually. The query set should mix category, comparison, problem, and brand queries — the actual questions a potential customer would ask an AI.

How often should I run a GEO audit?

Run a full audit quarterly, with monthly light check-ins on your top 10 queries. Also run off-cycle audits after major content changes (re-audit affected pages 3-4 weeks later) and after major AI platform updates. GEO auditing is recurring because AI models, retrieval systems, and competitors all change continuously.

What's the difference between a manual and automated GEO audit?

A manual audit (2-4 hours) gives qualitative depth — you read every AI response and develop intuition for your category. An automated audit (~60 seconds) gives scale, consistency, and scheduled re-measurement but less qualitative feel. Most teams run one manual audit to learn the methodology, then automate ongoing measurement.

What should a GEO audit tell me?

Two layers. The outcome: are you cited, for which queries, how prominently, versus which competitors. The diagnosis: which technical, content, and authority factors explain that outcome. A useful audit produces both a score and a prioritized fix list — a score without a diagnosis tells you that you have a problem but not how to fix it.

Can I run a GEO audit myself for free?

Yes. The seven-step manual methodology in this guide requires no paid tools — just time and access to the AI platforms. Alternatively, the LumenGEO free audit automates the core methodology and returns a scored diagnosis in about 60 seconds with no signup. Both are free; the manual version gives more depth, the automated version gives speed.

What's the first thing a GEO audit should check?

Technical crawler access (Step 2). Before measuring citations, confirm AI search engines can actually reach your content — check robots.txt for blocks on GPTBot, ChatGPT-User, PerplexityBot, and others, and confirm Bing indexation. A site that fails this has a hard ceiling on citation regardless of content quality, so checking it first prevents wasted effort diagnosing content when the real problem is access.

How do I turn a GEO audit into action?

Step 7 — prioritize fixes by impact. Order the diagnosed issues: citation-blocking fixes first (crawler access, indexation), then content extractability fixes on your highest-value pages, then authority and brand-mention work. Order also by how many queries each fix affects. The prioritized list is the audit's real deliverable — the score tells you where you stand, the list tells you what to do Monday morning.

Should I audit competitors too?

Yes — Step 4 is competitor analysis, and it is one of the most valuable parts of the audit. Seeing which competitors get cited, for which queries, and what content of theirs is being cited gives you a concrete roadmap of what AI search engines reward in your specific category. Competitors' cited content shows you the target far more precisely than generic best practices.