Content Audit for AI SEO: A Step-by-Step Process

What Is a Content Audit for AI SEO?

A content audit for AI SEO is a systematic review of your existing website content to determine how well it performs in AI-powered search environments. Unlike a traditional content audit that focuses on keyword rankings, traffic, and backlink profiles, an AI content audit evaluates whether AI models -- ChatGPT, Gemini, Perplexity, Claude, and Copilot -- can access, parse, and cite your pages when answering user questions.

The core question is simple: if someone asks an AI about your topic, will the AI use your content as a source?

Most businesses already have dozens or hundreds of published pages. The problem is that content written for Google's algorithm often fails in AI search. AI models don't rank pages -- they select sources. Content that lacks clear structure, schema markup, or citation-friendly formatting gets skipped entirely, regardless of how well it performs in traditional search.

If you're new to the concept of AI search optimization, start with our guide on what AI SEO is and why it matters. This article assumes you understand the basics and are ready to audit your existing content library.

Why Your Existing Content Needs an AI Audit

You've likely invested months or years building your content library. Before creating anything new, it's faster and more cost-effective to optimize what you already have. Here's why an AI audit should come first:

The visibility gap is enormous. 88% of pages cited by Google AI Mode are NOT in Google's top 10. Your top-ranking content may be completely invisible to AI. Conversely, buried pages with strong structure might already be getting AI citations you don't know about.

Content structure matters more than content quality in AI retrieval. A well-structured mediocre article will get cited over a brilliant article with poor formatting. AI models extract information from specific patterns -- BLUF summaries, quotable chunks, and clearly labeled sections. If your content doesn't match these patterns, AI skips it.

Schema markup is your competitive advantage. Research shows that FAQ Schema improves AI content interpretation from 16% to 54%. Most websites have minimal or no schema markup -- adding it to existing content is one of the highest-leverage changes you can make. Our JSON-LD basics guide covers the essential types.

You can measure the "before." Running an audit now gives you a baseline. When you make improvements, you can track exactly how your AI visibility changes -- page by page, platform by platform.

Step 1: Inventory All Pages

Before you can audit content, you need a complete list of what you have. This step creates your master spreadsheet.

How to build your page inventory

1. Crawl your site using Screaming Frog, Sitebulb, or Ahrefs Site Audit. Export all HTML pages (exclude images, CSS, JS, and PDFs for now).
2. Pull analytics data from Google Analytics 4 or your analytics tool. For each URL, capture: page views (last 90 days), average time on page, bounce rate, and top entry keywords.
3. Check your XML sitemap against your crawl. Pages missing from the sitemap may also be missing from AI indexes. See our XML sitemap best practices guide for details.
4. Categorize each page by type: blog post, product page, service page, landing page, about/team, legal, or other.
5. Tag business priority -- High (directly drives revenue or leads), Medium (supports sales or builds authority), Low (informational or outdated).

What your inventory spreadsheet should include

| Column | Description | |---|---| | URL | Full page URL | | Page title | H1 or title tag | | Content type | Blog, product, service, landing, etc. | | Word count | Total word count | | Business priority | High / Medium / Low | | Monthly traffic | Last 90 days average | | Last updated | Date of last meaningful edit | | AI audit score | Filled in during steps 2-5 |

Start with your High and Medium priority pages. A typical first audit covers 50-100 URLs. You can expand to the full site in subsequent quarters.

Step 2: Check AI Visibility Per Page

This step answers the most important question: is AI already citing this page?

Manual visibility check

For each page in your inventory, formulate 2-3 questions that the page should be able to answer. Then ask those questions to ChatGPT, Gemini, Perplexity, and Claude.

Example: If your page is about "best CRM for small businesses," ask:

• "What is the best CRM for small businesses?"
• "Which CRM should a startup with 5 employees use?"
• "Compare CRM tools for small teams"

For each query, record:

• Cited? Yes / No -- was your URL, brand, or data mentioned?
• Platform -- which AI model(s) cited you?
• Citation type -- direct link, brand mention, or data reference?
• Position -- where in the response did the citation appear (first paragraph, middle, end)?

Automated visibility tracking

Manual checks don't scale. For ongoing monitoring, tools like AImetrico track your AI citations across ChatGPT, Gemini, Perplexity, and Copilot automatically. You get a per-page visibility score and can track changes over time.

You can also check referral traffic in GA4: filter by source containing "chatgpt.com," "perplexity.ai," "claude.ai," or "copilot.microsoft.com" to see which pages already receive AI-driven visits.

Scoring for Step 2

• Cited by 3+ AI platforms: 25 points
• Cited by 1-2 platforms: 15 points
• Not cited but accessible to crawlers: 5 points
• Blocked or inaccessible: 0 points

Step 3: Evaluate Content Structure

AI models extract information in specific patterns. Content that matches these patterns gets cited; content that doesn't gets ignored. Review each page for the following structural elements:

BLUF (Bottom Line Up Front)

44.2% of AI citations come from the first 30% of content. Check whether each page leads with a direct answer to the primary question before elaborating with details, context, and examples. Pages that bury the answer after 1,500 words of introduction are nearly invisible to AI. Our BLUF principle guide explains how to restructure content effectively.

Quotable chunks

AI models extract 50-150 word fragments that stand alone as complete answers. Review your content for self-contained paragraphs that could be quoted directly. Content with clear quotable chunks gets 2.3x more citations than unstructured text.

Look for:

• Definition paragraphs -- "X is defined as..."
• Step summaries -- "To do X, follow these three steps..."
• Comparison statements -- "Unlike X, Y provides..."
• Data-backed claims -- "Research shows that X leads to Y..."

Heading structure

Check that each page uses a logical heading hierarchy (H1 > H2 > H3) with descriptive, question-based headings. AI models use headings to navigate content and identify relevant sections. Headings like "More Information" or "Section 3" tell AI nothing.

Lists and tables

Content in listicle format receives 74.2% of AI citations. Tables, numbered lists, and comparison formats are significantly easier for AI to parse and quote than wall-of-text paragraphs.

Scoring for Step 3

• BLUF present and effective: 10 points
• Quotable chunks (3+ per page): 10 points
• Clear heading hierarchy: 5 points
• Lists/tables for key information: 5 points
• Maximum for Step 3: 30 points

For detailed guidance on structuring content for AI citation, see writing content that AI models want to cite.

Step 4: Check Schema Markup

Structured data helps AI models understand what your content is about without guessing. Check each page for the following JSON-LD schema types:

Essential schema types to verify

| Schema Type | When to Use | AI Impact | |---|---|---| | Article / TechArticle | Every blog post and guide | Tells AI the content type, author, dates, and topic | | FAQPage | Any page with Q&A sections | Improves AI interpretation from 16% to 54% | | Organization | Homepage, about page | Establishes entity identity for brand mentions | | BreadcrumbList | All pages | Helps AI understand site hierarchy and context | | HowTo | Tutorial and process pages | Enables step extraction for procedural queries | | Product / Service | Commercial pages | Provides structured product data for recommendation queries |

How to check

1. Google Rich Results Test (search.google.com/test/rich-results) -- paste each URL and verify schema is valid and detected.
2. View page source -- search for application/ld+json to see what schema exists.
3. Schema Markup Validator (validator.schema.org) -- check for errors and warnings.

Common schema problems

• Missing entirely -- the most common issue. No schema means AI has to infer everything.
• Outdated dates -- dateModified hasn't been updated in years, signaling stale content.
• Missing author information -- no author name, credentials, or URL.
• Incomplete Organization schema -- missing logo, sameAs links, or contact info.

Scoring for Step 4

• Article + FAQ schema present and valid: 10 points
• Only Article schema: 5 points
• Schema present but has errors: 3 points
• No schema markup: 0 points

Step 5: Assess E-E-A-T Signals

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) signals influence whether AI models treat your content as a credible source worth citing. Review each page for:

Author attribution

• Is there a named author (not just "Admin" or "Staff")?
• Does the author have a bio with relevant credentials?
• Does the author byline link to an author page with more detail?
• Is the author's name consistent with their LinkedIn, social profiles, and other publications?

Publication and update dates

• Is a publication date visible on the page?
• Is an "Updated:" date shown when the content has been revised?
• Do datePublished and dateModified in schema markup match the visible dates?
• AI models penalize undated content -- it signals low confidence in timeliness.

Source citations

• Does the content cite external sources for data claims?
• Are links provided to studies, reports, or authoritative references?
• AI models trust content that shows its work -- unsourced claims get deprioritized.

Trust indicators

• HTTPS (not HTTP)
• Clear privacy policy and terms of service
• Physical address or contact information
• Industry certifications or awards mentioned

Scoring for Step 5

• Named author with credentials: 5 points
• Visible publication + update dates: 5 points
• External source citations (3+): 5 points
• Trust indicators present: 5 points
• Maximum for Step 5: 20 points

For a complete reference on what to check in your next audit cycle, see the AI SEO Checklist for 2026.

Step 6: Prioritize Fixes

Now you have scores for each page. The final step is turning data into an action plan.

Prioritization matrix

Sort your pages into four quadrants based on two factors: business priority (how important is this page to revenue?) and AI readiness score (how much work does it need?).

| | High Business Priority | Low Business Priority | |---|---|---| | Low AI Score (0-40) | Fix first -- high impact, needs work | Fix later or consolidate | | High AI Score (60+) | Monitor and maintain | Deprioritize |

Pages in the "High Priority + Low Score" quadrant are your biggest opportunities. These are pages that matter to your business but are currently invisible to AI.

Creating your fix queue

For each page that needs work, list specific actions:

1. Add BLUF summary (if missing) -- 30 minutes per page
2. Add FAQ section with schema -- 1 hour per page
3. Restructure into quotable chunks -- 1-2 hours per page
4. Add/fix schema markup -- 30 minutes per page
5. Add author bio and source citations -- 30 minutes per page
6. Full content rewrite for AI structure -- 4-8 hours per page

Assign each fix to a sprint or weekly task list. Aim to audit and fix 10-15 pages per week for a team of one.

AI Readiness Scoring Template

Here is the complete scoring template combining all six steps. Each page receives a score from 0 to 100.

| Audit Area | Criteria | Points | |---|---|---| | AI Visibility (Step 2) | Cited by 3+ platforms | 25 | | | Cited by 1-2 platforms | 15 | | | Accessible but not cited | 5 | | | Blocked/inaccessible | 0 | | Content Structure (Step 3) | BLUF present | 10 | | | 3+ quotable chunks | 10 | | | Clear heading hierarchy | 5 | | | Lists/tables for key info | 5 | | Schema Markup (Step 4) | Article + FAQ schema valid | 10 | | | Only Article schema | 5 | | | Schema with errors | 3 | | | No schema | 0 | | E-E-A-T Signals (Step 5) | Named author + credentials | 5 | | | Visible dates | 5 | | | External citations (3+) | 5 | | | Trust indicators | 5 | | TOTAL | | 100 |

Score interpretation:

• 70-100: AI-ready. Monitor and maintain.
• 40-69: Needs targeted improvements. Prioritize based on business value.
• 0-39: Significant restructuring required. These pages are effectively invisible to AI.

The average unoptimized website page scores around 35. After a focused audit and remediation cycle, most pages can reach 65-80 within 4-6 weeks.

Quick Wins vs Long-Term Improvements

Not all fixes take the same effort. Separate your action items into quick wins (under 1 hour per page) and long-term improvements (multiple hours or structural changes).

Quick wins (implement this week)

• Add a BLUF summary to the top of every article -- a 3-5 sentence "Key Takeaways" box that directly answers the page's main question. This alone can improve citation rates within days.
• Add an FAQ section with 4-6 questions to your top pages. Include FAQPage schema markup. FAQ content is cited disproportionately by AI models.
• Update dateModified in schema and add a visible "Updated:" date on pages you've recently revised. Freshness signals matter.
• Fix robots.txt if AI crawlers are blocked. This is the single highest-impact technical fix -- a page that AI can't access will never be cited.
• Add missing H2/H3 headings to break up long sections. Use question-format headings where possible.

Long-term improvements (plan for next 4-8 weeks)

• Restructure content into quotable chunks -- rewrite paragraphs so each 50-150 word section can stand alone as a complete answer.
• Add comprehensive schema markup across all page types -- Article, Organization, FAQ, HowTo, Product as applicable.
• Build author pages with credentials, publications, and expertise signals for every content contributor.
• Add original data and research to differentiate your content. AI cites sources with unique information gain over those that repeat common knowledge.
• Implement a content refresh calendar -- schedule quarterly reviews to keep content current. See our guide on updating old content for AI for the complete process.

Tools for AI Content Auditing

The right tools make your audit faster and more accurate. Here's what to use at each stage:

Page inventory and crawling

• Screaming Frog -- crawls your site and exports all URLs with metadata (word count, schema, status codes, headings). Free for up to 500 URLs.
• Sitebulb -- visual site audit with content quality scores and internal linking analysis.
• Ahrefs Site Audit -- combines crawl data with keyword rankings and backlink metrics.

AI visibility checking

• AImetrico -- automated AI visibility scoring across ChatGPT, Gemini, Perplexity, and Copilot. Tracks citations over time.
• Manual testing -- ask each AI platform directly. Time-consuming but free.
• GA4 referral reports -- filter traffic sources to identify pages already receiving AI-driven visits.

Schema validation

• Google Rich Results Test -- validates JSON-LD and shows which rich result types are detected.
• Schema Markup Validator (validator.schema.org) -- detailed error and warning reporting.
• Merkle Schema Generator -- helps create JSON-LD from scratch if you don't have schema at all.

Content structure analysis

• Hemingway Editor -- identifies complex sentences, passive voice, and readability issues.
• Surfer SEO -- content structure analysis with NLP term recommendations.
• Manual review -- nothing replaces reading your own content and asking: "Could an AI quote this paragraph as a standalone answer?"

Ongoing monitoring

After the initial audit, set up recurring checks. Monthly spot-checks on your top 20 pages take about 2 hours and catch regressions early. Full quarterly audits keep your entire content library AI-optimized.