What Our AI Search Audit Actually Checks (Full Methodology)

You submit your URL. We send you a report. But what happens between those two steps?

Most agencies treat their audit process like a trade secret. We think that’s backwards. If you understand what we check and why, you can decide whether an audit makes sense for your business. Or skip us entirely and do it yourself. We’d rather be transparent and lose a few leads than send vague promises to people who don’t know what they’re buying.

Here’s every step of our AI search audit, why each one matters, and what we actually look at.

Step 1: AI citation check across four platforms

This is the core of the audit. We query ChatGPT, Perplexity, Claude, and Google’s AI Overviews with questions your customers would ask about your industry.

Not generic questions. Specific ones. If you’re a CRM company, we’re not asking “what is a CRM?” We’re asking “what’s the best CRM for mid-market B2B sales teams with Salesforce migration needs?” The kind of questions that trigger real recommendations.

For each query, we record:

• Whether your brand gets mentioned
• What competitors get cited instead
• The exact phrasing AI uses to describe each brand
• Which sources AI links to (your site, review platforms, competitors’ pages)
• Whether the mention is positive, neutral, or a warning

Why four platforms? Because each one has different source preferences. Research from Detailed.com shows ChatGPT pulls 47.9% of its top citations from Wikipedia. Perplexity gets 46.7% of its top citations from Reddit. Google AI Overviews spreads citations more evenly across Reddit, YouTube, Quora, and LinkedIn. Being visible on one platform means nothing if you’re invisible on the other three.

We run between 15 and 30 queries per audit, depending on how broad your industry is. A single-product SaaS company needs fewer queries than a marketing agency with six service lines.

What this tells you

Most businesses have never asked an AI engine about their own industry. The gap between “we think we’re visible” and reality is usually painful. We’ve seen companies with $500K annual marketing budgets that don’t show up in a single AI response for their primary category.

The citation check gives you a baseline. You can’t improve what you haven’t measured.

Step 2: Technical AEO score

AI engines can only cite content they can access and understand. Sounds obvious. But roughly half the sites we audit have at least one technical issue blocking AI visibility.

Here’s the full technical checklist:

Crawler access

We check your robots.txt file for blocks against AI-specific crawlers. The ones that matter most:

• GPTBot (OpenAI/ChatGPT)
• CCBot (Common Crawl, used by multiple AI systems)
• Claude-Web (Anthropic)
• Google-Extended (Gemini training data)

Blocking GPTBot means ChatGPT can’t reference your content. We’ve audited three different B2B SaaS companies in the past four months that had Disallow: / rules for GPTBot. Their dev teams added these blocks without understanding the trade-off. These were fixable in five minutes, and visibility improved within weeks.

We also check for indirect blockers: login walls, paywalls, JavaScript-only navigation that crawlers can’t follow, broken canonical tags, and slow server response times. Any of these can silently prevent AI from reading your pages.

Schema markup

We run your site through a structured data audit looking for:

• Organization schema on your homepage with sameAs links to LinkedIn, Crunchbase, social profiles
• FAQ schema on relevant pages (this feeds AI Q&A directly)
• Article schema on blog posts
• Person schema for author pages
• Service/Product schema mapping what you actually offer
• HowTo schema on instructional content

Schema markup is how AI engines read structure rather than guessing it from paragraphs. A site with proper Organization + FAQ + Service schema gives AI five times more structured data points to work with than a site with nothing. That’s the difference between being “a website about CRM” and being “Acme CRM, a mid-market B2B CRM platform founded in 2019 with integrations for Salesforce, HubSpot, and Slack.” You can generate schema markup for free with our tool if you want to get started before the audit.

Content structure

AI models prefer content they can extract in chunks. We evaluate your top pages for:

• Question-based H2/H3 headings that mirror what users actually ask
• Direct answers in the first 40 to 60 words after each heading (the “answer capsule”)
• Specific statistics with proper source attribution
• Standalone sections that make sense without reading the full page

According to research from Backlinko, 72.4% of content cited by ChatGPT includes a concise answer capsule right after the heading. If your content buries the answer three paragraphs into each section, AI will pull from a competitor who leads with the answer.

llms.txt

We check whether you have an llms.txt file (the AI equivalent of robots.txt). This file tells language models what your site is about, which pages matter most, and how to categorize your brand. Most sites still don’t have one. If you’re in a competitive space, adding one is a low-effort advantage.

Content freshness

AI engines show a strong recency bias. As Metehan Yesilyurt from AEO Vision put it on LinkedIn: “ChatGPT prioritizes RECENT over PERFECT. That amazing guide from 2022? It’s losing to mediocre content published yesterday.” We flag any high-value pages that haven’t been updated in over 12 months.

Step 3: Competitor comparison

Knowing your own visibility means little without context. We benchmark you against your top three competitors.

For each competitor, we map:

• How often they appear in AI responses for your target queries
• What language AI uses when recommending them
• Which platforms they’re strongest on (some brands dominate on Perplexity but are invisible on ChatGPT)
• What content formats AI cites from their sites
• Their schema markup status
• Their crawler access settings

The competitor analysis often surfaces the most actionable insights. When a competitor gets cited for a query and you don’t, we can usually trace it back to a specific difference: they have FAQ schema, you don’t. They published a data study on the topic, you have a generic blog post. They’re on Reddit and YouTube for that topic, you’re not.

The competitive gap tells you what to prioritize. If every competitor already has strong schema and you don’t, that’s your first fix. If your content quality is better but your entity architecture is weaker, that changes the plan.

Step 4: Prioritized action plan

Every audit ends with a ranked list of fixes. Not a 40-page PDF you’ll never read. A specific list ordered by impact and effort.

Typical high-priority items include:

• Quick wins (under 2 hours): Unblock AI crawlers in robots.txt. Add llms.txt. Fix broken schema.
• Medium effort (1 to 2 weeks): Add FAQ schema to top 10 pages. Restructure headings for answer capsule format. Update stale content.
• Strategic plays (1 to 3 months): Build cross-platform entity presence. Create original research AI can cite. Develop structured content hubs around target queries.

Each item includes what to do, why it matters for AI visibility, and how to measure whether it worked. We include GA4 setup instructions for tracking AI referral traffic using custom channel groups, because if you can’t see AI-driven visits separately from organic, you can’t measure progress.

You can take the plan and execute it yourself. That’s fine. Or you can ask us to handle it. The audit is designed to be useful either way.

Why each platform gets its own analysis

One mistake we see from other agencies offering “AI audits” is treating AI search as one thing. It’s not. Each platform has a distinct personality, and what works on one might fail on another.

ChatGPT

Favors authoritative, encyclopedic sources. Wikipedia accounts for 7.8% of all ChatGPT citations. Your best bet: well-sourced content with specific statistics, strong domain authority, and clear author credentials. ChatGPT also shows a strong recency bias, so stale content gets deprioritized fast.

Perplexity

Community-driven. Reddit accounts for 6.6% of all Perplexity citations and 46.7% of its top 10 sources. If your brand is discussed on Reddit, you’re in a strong position on Perplexity. If you only exist on your own website and LinkedIn, Perplexity will probably recommend someone who has Reddit presence.

Google AI Overviews

Balanced approach. Cites Reddit (2.2%), YouTube (1.9%), Quora (1.5%), and LinkedIn (1.3%) more evenly. Google also leans into its Shopping Graph for product queries, pulling from 35 billion product listings. The Pew Research Center found that government (.gov) sites get cited 6% of the time in AI Overviews, compared to just 2% in standard search results. Authority signals matter more here.

Claude

Less studied than the other three, but in our audit experience, Claude favors well-structured technical content with clear sourcing. We include it because Anthropic’s partnership with Apple means Claude’s market share is growing, and brands that ignore it now will scramble later.

What the audit costs (nothing)

The full AI visibility audit is free. We run it because once you see the data, the path forward becomes obvious. Most businesses don’t know AI engines are answering questions about their industry right now. When they see competitors getting cited and they’re not, the problem stops being theoretical.

We don’t gate the insights behind a sales call. You get the report. You decide what to do with it. If you want our team to fix the issues, we’re here. If you want to handle it internally, the action plan tells you exactly how.

What an audit won’t tell you

Transparency goes both ways. Here’s what our audit doesn’t cover:

• Exact AI ranking algorithms. Nobody knows exactly how these models weight sources. We observe patterns and base recommendations on what works, not on insider knowledge.
• Guaranteed citation timelines. AI training data updates happen on each platform’s schedule, not ours. Technical fixes like unblocking crawlers can show results in weeks. Content strategy changes take 3 to 6 months to reflect in citations.
• Traditional SEO issues. We focus on AI visibility specifically. If you need a full SEO audit, that’s a different engagement.

Anyone promising you’ll be “cited by ChatGPT in 30 days” is selling something they can’t deliver. We tell you what’s broken, how to fix it, and what timeline is realistic.

Who should get an audit

Be honest with yourself here. An AI search audit makes sense if:

• Your business depends on being found when people ask questions about your industry
• You’ve noticed competitor brands showing up in AI responses where you’re absent
• You’ve invested in content marketing but haven’t adapted for AI search
• You don’t know whether AI crawlers can even access your site

If your business runs on referrals and word-of-mouth and search has never mattered, this probably isn’t urgent for you. But if search drives revenue, AI is eating into that channel. Ahrefs found that position-one click-through rates for informational keywords dropped from 7.6% to 3.9% between December 2023 and December 2025, largely because AI Overviews answer the question before anyone clicks.

Check your own visibility first with our free AI visibility checker. If the results surprise you, an audit will tell you exactly why.