How to Do a GEO Audit

Most B2B SaaS companies have spent years building their SEO presence, but buyers are shifting. They’re asking ChatGPT, Perplexity, Gemini, and Claude to recommend software, compare vendors, and shortlist solutions before they ever visit a website. If your company isn’t showing up in those AI-generated answers, you’re invisible during the research phase that matters most. A GEO audit tells you exactly where you stand. It’s a structured review of how generative AI tools currently describe your company, which competitors they cite instead, and what you need to fix to earn your place in those answers. This isn’t about chasing a trend. It’s about understanding a new channel that’s already influencing your pipeline, and working out what to do about it.

The process isn’t complicated, but it does require discipline. You need to query the right tools, record what they say, compare it against your competitors, and then prioritise fixes based on what will actually move the needle. We’ve broken it into five steps below, with enough detail that your team can run one this week.

What is a GEO audit?

A GEO audit is a systematic check of how generative AI tools represent your brand, your category, and your competitors. GEO stands for Generative Engine Optimisation, and it’s the practice of making sure AI models can find, understand, and accurately cite your company when buyers ask relevant questions. The audit is the diagnostic step: it tells you what’s working, what’s broken, and where the gaps are.

Strong SEO foundations are an input to GEO, not a replacement for it. Your rankings in Google still matter. But GEO adds a layer of work on how AI models learn about your company: the sources they pull from, how consistently your positioning is described across the web, and whether you’re cited in the right contexts. AI tools increasingly form a fourth pillar of digital visibility alongside organic, paid, and social channels.

Think of a GEO audit as the equivalent of a technical SEO audit, but for AI search. Where a technical audit checks crawlability, indexation, and site health, a GEO audit checks whether AI tools can build an accurate, credible picture of who you are and what you do. If they can’t, you won’t appear in the answers your buyers are reading.

What a GEO audit covers

A thorough audit touches several areas, and each one feeds into the next. You’re not just looking at one thing; you’re building a complete picture of your AI visibility.

The core areas include:

• Brand representation: how AI tools describe your company when asked directly, and whether those descriptions match your actual positioning.
• Category presence: whether your company appears when buyers ask about your software category (e.g., “best endpoint detection tools for mid-market” or “top subscription billing platforms”).
• Competitor citations: which companies AI tools recommend instead of yours, and what sources they draw from.
• Content gaps: topics and questions where your company has no content for AI tools to reference.
• Structured data: whether your site’s schema markup gives AI models the signals they need to understand your product, pricing model, and company details.
• Source authority: which third-party sites, reviews, and publications AI tools pull from when constructing answers about your category.

A good audit checklist for agencies will cover all of these in a structured, repeatable format. The goal is to leave nothing to guesswork.

Step 1: Check how AI tools describe your company

Start by querying ChatGPT, Perplexity, Gemini, and Claude with the questions your buyers would ask. Ask each tool to describe your company. Ask it to recommend software in your category. Ask it to compare you against a specific competitor. Record every response.

You’re looking for three things. First, accuracy: does the AI get your product right, or does it describe features you don’t have, or miss the ones you do? Second, completeness: does it mention your target market, your pricing model, your key differentiators? Third, consistency: do all four tools tell roughly the same story, or does one describe you as an enterprise platform while another calls you a startup tool?

This step usually surfaces some uncomfortable truths. We’ve seen companies described as offering products they discontinued two years ago, or positioned in categories they don’t compete in. The AI models are only as good as the data they’ve ingested, and if your web presence sends mixed signals, the output will be mixed too. Run at least 15 to 20 queries across all four tools. Use variations: different phrasings, different buyer personas, different stages of the buying journey. Document everything in a spreadsheet with columns for the tool, the query, the response, and your assessment of accuracy.

Step 2: Find where competitors are cited instead of you

This is where the audit gets competitive. For every category-level query where your company should appear but doesn’t, note which competitors do appear. Then dig into why.

AI tools cite sources. Perplexity shows them explicitly. ChatGPT with browsing shows them too. Look at the sources being referenced for your competitors and ask: do they have a presence on sites where you don’t? Are they mentioned in analyst reports, review platforms, or industry publications that you’re absent from? Do they have more consistent third-party descriptions of what they do?

Build a competitor citation matrix. List the top five to eight competitors across the top of a spreadsheet, and list the queries down the side. Mark which competitors appear in each answer. Patterns will emerge quickly. You might find that one competitor dominates because they have strong G2 and Capterra profiles with recent reviews. Another might appear because they’ve been cited in several industry roundups.

This step is where you start to see the gap between what AI tools synthesise and what your company actually offers. The fix isn’t always about creating more content. Sometimes it’s about getting your existing positioning into the right places.

Step 3: Map your content and entity gaps

Now you know what AI tools say about you and where competitors outperform you. The next step is to map the content gaps that explain the difference.

Pull up the queries where you’re absent or inaccurately described. For each one, check whether you have content on your site that directly answers that question. Not a vague blog post that touches on the topic, but a clear, well-structured page that an AI model could confidently pull from.

Topic clusters matter here. AI tools don’t just look at individual pages; they build understanding from patterns across your site. If you claim to be a leader in, say, API security, but you have two blog posts and no dedicated product page on the topic, the AI model has very little to work with. Gripped builds content architecture around the questions buyers ask AI tools, closing the gap between what a company says and what AI tools can actually find and verify.

Entity gaps are the other half. An entity, in this context, is a recognisable “thing” that AI models can identify: your company name, your product names, your founders, your category. If your entity signals are weak or inconsistent across the web, AI tools struggle to connect the dots. Check whether your company is described consistently on LinkedIn, Crunchbase, G2, your own site, and any press coverage. Inconsistency here is one of the most common reasons companies get left out of AI answers.

Step 4: Check your structured data

Structured data is the technical layer that helps AI models understand what your pages contain. If your site lacks proper schema markup, you’re making it harder for generative engines to parse your content accurately.

At minimum, your site should have Organisation schema with your company name, logo, URL, and founding date. Product schema should cover your core offerings, including pricing model where applicable. FAQ schema on relevant pages helps AI tools pull specific answers. If you publish case studies, Article schema with author and date information adds credibility signals.

Run your key pages through Google’s Rich Results Test and Schema.org’s validator. Look for errors, warnings, and missing fields. Pay particular attention to whether your product pages have schema that reflects what you actually sell. A step-by-step guide to GEO preparation will typically flag structured data as one of the highest-impact, lowest-effort fixes available.

For SaaS companies, there’s also a technical performance angle. Sites that meet Core Web Vitals thresholds and score above 90 on Lighthouse tend to get crawled and indexed more reliably, which feeds into the data AI models can access. Don’t overlook the basics.

Step 5: Prioritise the fixes

You’ll come out of the first four steps with a long list of issues. Not all of them matter equally. The final step is to sort them by impact and effort, so your team works on the right things first.

Group your findings into three tiers:

1. High impact, low effort: fixing inaccurate structured data, updating your company description on third-party profiles, correcting outdated information on your own site. These can often be done in a week.
2. High impact, high effort: creating new content clusters for topics where you’re absent, building a review generation programme on G2 or Capterra, earning citations in industry publications. These take weeks or months but move the needle significantly.
3. Low impact: minor inconsistencies that don’t affect buyer-facing queries, or gaps in categories you don’t actively compete in.

Teams that treat this like a sprint rather than a one-off project get better results. Gripped runs 30-day sprints with real-time reporting, which works well for GEO because you can re-query the AI tools after each round of fixes and measure whether your visibility has changed. The feedback loop matters. If you fix your structured data and update your G2 profile in week one, you can check whether Perplexity’s answers have shifted by week four.

What a good audit output looks like

A useful GEO audit deliverable isn’t a 60-page PDF that sits in a drawer. It’s a working document that your marketing and content teams can act on immediately.

The best outputs we’ve seen include a summary dashboard showing current AI visibility scores across tools, a competitor citation matrix, a prioritised list of fixes with owners and deadlines, and a set of benchmark queries you’ll re-run monthly to track progress. Some teams also include a “source gap” analysis showing which third-party sites they need to build a presence on.

The format matters less than the clarity. Every recommendation should have a clear owner, a deadline, and a way to measure whether it worked. If your audit just says “improve content” without specifying which topics, which formats, and which AI tools you’re targeting, it’s not actionable enough.

A practical framework for auditing AI search readiness will tie every finding back to a specific query, a specific AI tool, and a specific fix. That’s the standard to aim for.

Common questions

What tools do you need for a GEO audit?

You need access to ChatGPT (preferably with browsing enabled), Perplexity, Gemini, and Claude. These are the four AI tools that B2B buyers most commonly use for vendor research. Beyond that, you’ll need Google’s Rich Results Test for structured data validation, and a spreadsheet to track queries, responses, and competitor citations. Some teams use paid tools like Semrush or Ahrefs to cross-reference which of their pages are being cited as sources, but these aren’t strictly necessary for the audit itself. The most important tool is discipline: running enough queries across enough tools to get a reliable picture, not just cherry-picking a few prompts that make you look good.

How often should you run a GEO audit?

Run a full audit quarterly. AI models update their training data and retrieval sources regularly, so what’s true today may not be true in three months. Between full audits, run a lighter monthly check using your benchmark queries to spot any significant changes. If you launch a new product, enter a new category, or a major competitor makes a big move, run an ad hoc audit on those specific areas. The pace of AI adoption in search means that waiting six months between audits risks falling behind without realising it.

Getting started

Running a GEO audit isn’t optional for SaaS companies that rely on inbound pipeline. Buyers are already using AI tools to research your category, and the companies that show up in those answers have a real advantage in the consideration phase. The five steps above give you a repeatable process: check your current AI representation, map competitor citations, identify content and entity gaps, validate your structured data, and prioritise fixes by impact.

The companies that treat GEO as an ongoing programme rather than a one-off project will compound their advantage over time. Each fix improves the data AI models can access, which improves your visibility, which feeds back into more accurate citations.

If you’re a SaaS or tech company looking for help connecting your SEO, content, and GEO work into a single growth programme, Gripped works exclusively with B2B SaaS and tech businesses and can help you turn audit findings into pipeline. Get your free growth audit to see where you stand.