I get asked some version of this question in almost every conversation I have about AEO: ‘What do you actually find in the audit?’
It’s a fair question. AEO is abstract enough that even people who understand the concept aren’t sure what a diagnosis looks like in practice. So let me walk you through a representative audit — drawing from the patterns I see most consistently — and show you what the problems actually look like on a real site.
The site in this example is a service business in a competitive professional services category. Solid brand, good content, decent SEO fundamentals. And almost completely invisible to AI search.
Starting Point: The AI Visibility Check
Before any technical analysis, I run what I call the AI Visibility Check: I go to ChatGPT, Perplexity, and Google Gemini and ask the questions this site’s ideal customer would ask.
‘What is the best [service type] for [use case]?’ ‘How does [specific process] work?’ ‘What should I look for in a [provider type]?’
The site doesn’t appear in a single response across three platforms. The competitors being cited are sites with weaker content and lower domain authority. The Retrieval Gap is real and measurable before I’ve looked at a single line of code.
Finding 1: No FAQ Schema Anywhere on the Site
The site has a genuine FAQ page. It’s well-written. The questions are relevant. The answers are clear. And it has zero schema markup.
From an AI extraction standpoint, this FAQ page is invisible. The AI can’t identify it as a question-answer resource. It’s just unstructured text that happens to be formatted like questions and answers for human readers.
Adding FAQPage JSON-LD to this page — with questions rewritten to match actual AI prompts — is the single highest-impact fix in the audit. It doesn’t require new content. It requires wrapping existing content in the schema the AI needs to use it.
Finding 2: Answer Blocks That Are Too Long
The site’s service pages are well-written from an SEO perspective — thorough, authoritative, well-structured. They’re also structured in a way that buries the direct answer inside extended explanations.
A typical section opens with two paragraphs of context before stating the actual answer to the implicit question. Those context paragraphs are the problem. AI extraction is scanning for the direct answer, and it’s finding qualifying language instead.
The fix is surgical: add a two-sentence answer block at the top of each major section. Keep the existing content. Just lead with the answer before explaining it.
Finding 3: Entity Inconsistency Across Platforms
The Organization schema on the site lists the business name one way. The Google Business Profile uses a slightly different version. LinkedIn uses a third variation. The address format differs between the schema and GBP.
Each of these individually seems trivial. Collectively, they create an entity trust problem. AI systems evaluating whether to cite this business as an authoritative source are seeing inconsistent signals about who and where this entity is. Inconsistency reads as uncertainty, and uncertain sources don’t get cited.
Entity normalization — making every instance of the brand’s identity consistent across every platform where it appears — is unglamorous work. It’s also essential AEO hygiene.
Finding 4: Content That Talks Around the Question
Several pages on the site are structured around the company’s narrative rather than the customer’s question. The About page explains the company’s history. The Services page describes the team’s approach. The blog posts build case studies around outcomes.
These are legitimate content formats. They serve human readers well. They serve AI systems poorly, because none of them are structured as direct answers to specific questions.
The fix isn’t to delete this content. It’s to add question-first framing: lead each section with the question it answers, then deliver the answer, then provide the supporting narrative. The content becomes dually useful — readable for humans, extractable for AI.
Finding 5: No Freshness Signals
The site’s best content was published two to three years ago and hasn’t been updated. From an AI system’s perspective, content with no recent publication or modification date is a weak signal.
The content is still accurate. But the staleness is visible in the schema (no dateModified) and in the absence of any recent topical coverage that would signal active editorial engagement.
A quarterly review and update cycle — even minor updates with a refreshed dateModified — is part of a sustainable AEO content strategy. It’s not about manufacturing freshness. It’s about maintaining the recency signals that AI systems use to evaluate whether content is still the best available answer.
What the Audit Doesn’t Do
An AEO audit is a diagnostic, not a prescription. The audit tells you what’s broken and where the visibility gaps are. It doesn’t hand you a universal fix list, because AEO problems are site-specific. The solutions for a professional services firm are different from the solutions for a Shopify store or a news publication.
What the audit gives you is clarity: the exact problems, ranked by impact, with enough context to act on them — whether you do that work yourself, hand it to your internal team, or bring in someone who specializes in AEO implementation.
That’s what a $450 audit buys you. Not a strategy. A diagnosis. Clear, honest, and specific enough to be actionable.