All posts
GuidesMarch 5, 20259 min read

How to Do a GEO Audit: Step-by-Step Guide

Learn how to measure your brand's visibility in AI engines yourself. A professional GEO audit methodology explained step by step.

How do you measure your brand's visibility in ChatGPT, Claude, Gemini, and Perplexity? Most companies approach this question with assumptions — "we're probably fine" or "nobody's searching for us there." Both are dangerous. A GEO audit is the only way to establish an objective baseline.

What Is a GEO Audit?

A GEO audit is a systematic analysis of how your brand is represented across generative AI engines. It can be done manually — with real limitations — or professionally, with broader query sets, competitor benchmarking, and statistical confidence. This guide walks you through the steps to conduct a basic manual audit for your brand.

Step 1: Select Representative Queries (10-15 Questions)

The quality of your audit depends entirely on your query selection. A strong query set covers these categories:

  • Category queries: "What's the best [your service] for [target customer type]?"
  • Problem-solution queries: "I need [specific outcome] — who should I work with?"
  • Comparison queries: "What's the difference between [your brand] and [competitor]?"
  • Direct brand queries: "What do you know about [brand name]?"

Select 3-4 queries per category. Focus on questions your actual customers ask — think from the buyer's perspective, not your marketing department's.

Step 2: Systematic Testing Across 4 Engines

Run each query through ChatGPT, Claude, Gemini, and Perplexity in sequence. For each test:

  • Use the exact same question — do not paraphrase between engines.
  • Copy the complete response, not just the first few lines.
  • Note the time — some engines use real-time data that varies by session.
  • On Perplexity, record which sources are cited.

Step 3: Record and Score Responses

Evaluate each response against four criteria on a 0-3 scale:

  • Mention (0/1): Is your brand referenced at all?
  • Position (1-3): First mention, middle of list, or last?
  • Accuracy (0-3): Is the description correct? Right category? Right features?
  • Sentiment (0-3): Positive, neutral, or negative positioning?

Enter this into a spreadsheet: queries as rows, engines as columns, total score per cell (max 10). This table is the foundation of your cross-engine comparison.

Step 4: Competitor Benchmarking

Run the same query set for your top competitors. The goal: compare average scores across engines. Where a competitor outperforms you, analyze why:

  • Do they have a Wikipedia or Wikidata page?
  • Are they featured more frequently in industry publications?
  • Is their schema markup more comprehensive?
  • If they dominate a specific engine, what does that engine prioritize?

Step 5: Pattern Recognition

With all responses in front of you, look for patterns. Ask yourself:

  • Which signals are working? (When you are cited, what content type is the likely source?)
  • Which engine shows your weakest performance, and why?
  • Is your brand described consistently, or does the description shift across engines?
  • Which competitor dominates which query types?

Step 6: Interpret Your Results

Use your average score to assess your GEO position:

  • 7-10: Strong representation. Shift to optimization and ongoing monitoring.
  • 4-6: Mid-range. Gaps exist in specific engines or query categories. Prioritize accordingly.
  • 0-3: Critical gap. Start with foundational infrastructure: schema markup, Wikipedia presence, citation network.

The Limitations of Manual Audits

This guide gives you a solid starting point. But understanding its limitations is important:

  • 15 queries vs. 47 queries produce meaningfully different statistical confidence.
  • AI responses vary by session context, geographic location, and current model version.
  • Manual competitor coverage is typically limited to 2-3 rivals; professional audits cover 12 or more.
  • Identifying which specific signals drive which outcomes (causal analysis) is extremely difficult manually.

A professional GEO audit closes these gaps: 47 queries, 4 engines, 12+ competitors, statistical significance testing, and a prioritized 30-page action plan. When your manual audit surfaces inconsistencies or unexplained patterns, that's the signal that professional analysis is the right next step.

To start your brand's GEO audit, reach out through visivily.com. The initial assessment call is complimentary.

Curious about your brand's GEO score?

Start your audit