The GEO Content Audit: 20 Questions to Ask Before You Optimize Anything

A GEO content audit is a structured check of whether AI crawlers can reach your pages, whether your content is formatted for extraction, whether your claims are trusted enough to cite, and whether you're actually measuring any of it, run before you rewrite a single page. Skipping straight to optimization without this audit means guessing at fixes for a problem you haven't diagnosed. Contently's 2026 research found AI search visits grew 42.8% year over year, from 15.6 billion to 27.4 billion in Q1 2026 alone, while traditional Google search grew just 2.4% in the same period, which is why an outdated or incomplete audit now costs more than it used to.

This article gives you 20 specific questions across four audit areas: technical access, content structure, authority and trust signals, and measurement. Answer these honestly before you touch your content strategy.

Related read: for the fundamentals these questions are built on, see our full Generative Engine Optimization guide and our breakdown of how AI models decide which brands to mention.

• A GEO content audit differs from a traditional SEO audit because it asks whether AI systems can extract and cite a page, not just whether the page ranks.
• LLMs are 28 to 40% more likely to cite content with clear formatting such as hierarchical headings, bullet points, and tables, according to HubSpot's GEO research.
• The share of organic keywords triggering a Google AI Overview grew from about 1.5% to roughly 32% in the twelve months between September 2024 and September 2025, a twentyfold increase, per Onely's research.
• A widely cited Princeton and Georgia Tech study on generative engine optimization found that adding expert quotations lifted visibility by 41%, statistics by 33%, and citations by 28%, giving a rough ranking of which fixes matter most.
• Run this audit before optimizing anything, since fixing the wrong layer, content quality on a page AI crawlers can't even reach, wastes effort on a problem the audit would have caught first.

Before anything else, confirm that AI crawlers can technically access, render, and index your pages, since content quality is irrelevant if a crawler never sees it. This is the first audit layer and the one most teams skip, assuming that if Google can crawl a page, so can every AI system, which is not always true.

• Are GPTBot, ClaudeBot, PerplexityBot, and Google-Extended explicitly allowed in your robots.txt file, rather than blocked by a default or legacy rule?
• Do your priority pages return a clean response without relying on JavaScript rendering that some AI crawlers may not fully execute?
• Is your XML sitemap current, and does it include every page you actually want an AI system to find and cite?
• Do you maintain an llms.txt file that gives models a clear index of what your domain covers? Our llms.txt guide covers implementation.
• Are page load times and Core Web Vitals fast enough that a crawler completes a full page render rather than timing out partway through?

Once crawlers can reach your content, the next question is whether it's shaped so a model can extract a clean, quotable answer, which depends far more on structure than on length. HubSpot's GEO research found that LLMs are 28 to 40% more likely to cite content with clear formatting, including hierarchical headings, bullet points, numbered lists, and tables, than content without it.

• Does each priority page answer its core question in the first two to three sentences, before any narrative buildup?
• Are your headings phrased as the actual questions a buyer or researcher would ask, rather than generic keyword phrases?
• Do you use bullet lists, numbered steps, and comparison tables wherever you're describing options, steps, or trade-offs?
• Is each section self-contained enough to make sense on its own if a model quotes just that portion?
• Have you removed long storytelling intros that bury the direct answer below the fold?

Even well-formatted content won't be cited if its claims aren't specific, sourced, and corroborated elsewhere, since AI systems weigh trust signals as heavily as structure. The Princeton and Georgia Tech GEO study quantified this directly: adding expert quotations lifted visibility by 41%, adding statistics lifted it by 33%, improving fluency lifted it by 29%, and adding citations lifted it by 28%, giving a clear priority order for this layer of the audit.

• Does every major claim on the page carry a specific statistic, date, or named source instead of a vague generalization like "studies show"?
• Do you include expert quotations or named author attribution, rather than generic bylines like "admin" or "editorial team," which fail AI confidence checks regardless of content quality?
• Is your schema markup (Organization, Article, FAQPage, HowTo) implemented and validated, and does it match what's actually visible on the page?
• Do independent third-party sources, such as review platforms, press coverage, or industry publications, corroborate the same facts your own site states?
• Is your brand's entity information, including your About page, founder bios, and company facts, consistent across your site and external profiles like Wikipedia, LinkedIn, and Crunchbase?

An audit is only useful if you have a baseline and a repeatable way to check whether fixes worked, which is the layer most GEO efforts skip entirely. Without measurement, you can't tell whether a content change improved citation rate or whether any change was coincidental.

• Have you tested your actual target buyer questions directly in ChatGPT, Perplexity, Gemini, and Google AI Overviews to see whether and how your brand appears?
• Do you know which competitor domains are currently winning citations for your highest-priority queries, and why?
• Is your content on a defined refresh cadence, with citation rate tracked before and after each update? Freshness matters: pages updated within roughly the past two months earn a meaningful citation boost in multiple 2026 studies.
• Are you distinguishing between a citation, a linked source in an AI answer, and a mention, a brand named without a link, in your tracking? Our guide on tracking your brand across ChatGPT, Perplexity, and Gemini covers this distinction in more depth.
• Do you have a logged baseline AI visibility score, so future audits can measure whether work actually moved the needle rather than relying on impression?

Technical access should be checked first, since no amount of content quality matters if crawlers can't reach a page, followed by structure, then authority signals, then measurement, which validates whether the first three layers are working.

Repeat the full audit on a quarterly cadence, and re-run the technical and structure sections any time you publish significant new content, since AI platforms change citation behavior often and competitors are publishing continuously. Contently's guidance recommends logging baselines each round specifically so a team can see whether each cycle of work produced measurable improvement, rather than treating the audit as a one-time project.

How is a GEO content audit different from a standard SEO audit?

A traditional SEO audit asks whether a page ranks. A GEO content audit asks whether an AI system can access, extract, trust, and cite that same page, which depends on crawlability, structure, and sourcing in addition to the ranking factors SEO already covers.

Which of the four audit layers should we fix first?

Technical access first, since a page blocked from AI crawlers gets zero benefit from content improvements. After that, prioritize structure and authority fixes on your highest-intent pages rather than spreading effort across your entire site.

Do we need special tools to run this audit, or can it be done manually?

The technical and structure questions can be checked manually with your existing analytics and a robots.txt review. The measurement layer benefits from a dedicated AI citation tracking tool, since manually testing prompts across four or more platforms on a recurring basis doesn't scale well past a handful of pages.

How many pages should we audit at once?

Start with your ten highest-intent pages rather than your entire content library. Contently's research recommends prioritizing by impact, meaning the pages buyers already ask about and the pages competitors currently win citations on, over auditing by page count alone.

Don't try to answer all 20 questions across your entire site in one sitting. Pick your five highest-intent pages, work through the four audit layers in order, technical access, structure, authority, then measurement, and log where each page fails. That gives you a prioritized, evidence-based backlog instead of a guess about what to optimize first, and a real baseline to compare against once the fixes are live.