AI search audit: how to audit your AEO and GEO visibility, and when to get one done

TL;DR

• Traditional SEO audits miss AI retrieval entirely. Google ranks documents. LLMs retrieve semantically relevant passages and synthesize a single answer. Auditing for one does not cover the other.
• Three surfaces require three audit layers: web search (classic SEO), citations (real-time passage retrieval), and training data (offline brand associations).
• Citation rate is the new primary KPI. Measure the percentage of your priority buyer-intent queries that produce a brand mention in ChatGPT, Claude, Perplexity, and Gemini.
• Attribution is imperfect but trackable. UTM parameters, self-reported form fields, and CRM custom fields together build a defensible board slide.
• Initial citations can appear in 1-2 weeks with optimized content. A meaningful citation rate lift typically takes 3-4 months of consistent work.

Ahrefs data from early 2026 shows about 38% of AI Overview citations came from pages ranking in the top 10, so most of what AI cites is not what ranks. That gap is where pipeline goes invisible. Traditional SEO audits check backlinks, Core Web Vitals, and keyword rankings, but none of those signals tell you whether ChatGPT, Claude, or Perplexity will cite your brand when a buyer asks which tool to use. This guide explains what an AEO audit covers, how to run a basic one yourself, and how to decide whether a managed diagnostic is the right investment.

Core components of an AI visibility audit#

How AI audits differ from traditional SEO audits#

A traditional SEO audit measures domain authority, backlink profiles, Core Web Vitals, and keyword rankings. An AI visibility audit measures something different: whether your content can be extracted and cited by a large language model when a buyer asks a relevant question.

The core technical difference is the retrieval mechanism. Google scores and ranks full documents. LLMs use dense passage retrieval, a dual-encoder architecture that maps both questions and candidate passages into a shared vector space and selects the most semantically similar passages, not pages. Research by Karpukhin et al. found that dense retrievers achieve 9-19 percentage points higher top-20 accuracy than BM25 (a lexical keyword-matching algorithm). A page ranked first on Google is not automatically the passage an LLM extracts. Read the full breakdown of why traditional SEO audits miss AI visibility.

The three surfaces: web search, citations, and training data#

We view organic search through three surfaces, each requiring its own audit layer:

1. Web search: Classic SEO. Being discoverable when humans and AI agents search the web. Rankings, crawlability, and click-through rate matter here.
2. Citations: Satisfying LLMs at citation time, when a retrieval-augmented generation system pulls passages to build an answer. The CITABLE framework is designed specifically for this surface.
3. Training data: Brand associations surfaced without real-time retrieval. LLMs have prior beliefs about which brands exist in a category, formed during pre-training.

For a breakdown of what surfaces 2 and 3 require specifically, see what a GEO audit covers. A standard SEO audit covers surface one only. An AI visibility audit must cover all three. The foundations of SEO and AEO are largely shared, but the remaining gap is where competitive advantage lives in 2026.

Key KPIs for your AI search audit#

Traditional CTR and impressions are necessary but no longer sufficient as primary metrics. Buyers now research inside AI assistants without ever clicking through to your site, making impression-based metrics incomplete measures of buyer attention. The measurement stack for an AI search audit includes:

• Citation rate: The percentage of priority buyer-intent queries where a specific AI platform mentions your brand.
• Mention rate: How often your brand appears in AI answers, whether cited as a source or referenced in the synthesized response.
• Share of voice: Your brand's citation count as a percentage of total citations in your category across a defined query set.
• AI-referred MQLs: Leads where the first or last touch was an AI platform, tracked via UTM parameters and self-reported form fields.

For context on how these numbers behave in practice, our post on real citation rate benchmarks explains why platform-reported numbers frequently understate actual visibility.

Key metrics analyzed in AI search audits#

Measuring AI citation and mention rates#

To calculate your citation rate, define a query set of 100-200 buyer-intent questions your target customers ask. Prioritize queries by pipeline value (which searches historically drove closed deals) rather than search volume. Run each query across ChatGPT, Claude, Perplexity, and Gemini. Record whether your brand is mentioned, then divide mentions by total queries per platform to get a per-platform citation rate.

Doing this manually at scale is slow. Our AI visibility tracking tool automates this across hundreds of queries and six platforms, producing a share-of-voice breakdown against named competitors. Compare AI visibility audit tools if you're evaluating third-party options instead of a managed engagement. For teams evaluating third-party tools, our AI visibility platforms buyer's guide compares the leading options on measurement precision and CRM integration.

Verifying NAP consistency and brand mentions#

LLMs verify factual claims about your brand by checking consistency across independent sources. Google's AGREE research shows that claims appearing consistently across multiple independent sources receive higher confidence scores. This changes what off-page strategy means: it is no longer primarily about acquiring do-follow backlinks. It is about keeping the same accurate statement about your product live across Reddit, industry publications, comparison content, and your own site.

Check that your product descriptions, use cases, and positioning are consistent across your website, G2, Capterra, Reddit, and industry roundup articles. Discrepancies reduce LLM confidence in your brand's claims.

Optimizing for LLM passage retrieval#

Retrieval-augmented generation (RAG) systems extract passages, not pages. For your content to be a passage candidate, each section needs to independently answer one question within 200-400 words, use direct entity relationships in the copy (not just schema), and include verifiable facts with sources. Our CITABLE framework optimization workflows post covers how to automate this audit step.

Steps to evaluate your own AI citations#

Step 1: Audit brand citations in LLMs#

The starting point is direct brand queries. Ask your SEO specialist or content team to test "What is [your brand]?", "What does [your brand] do?", and "Is [your brand] good for [use case]?" across ChatGPT, Claude, Perplexity, and Gemini. The output you want from this step is a record of whether your brand is mentioned, how it is framed, and which competitors appear alongside it.

The same team should run competitor queries: "What are the best alternatives to [competitor]?" and "Compare [competitor A] vs [competitor B]." Absence from these responses is the first confirmed gap and the signal to act. Our Claude Code visibility audit guide shows how to automate competitor citation analysis at scale.

Step 2: Map your top 10 buyer-intent queries#

Select the 10 queries with the highest pipeline value, not the highest search volume. Work with your SEO manager to identify the questions your ICP asks when evaluating vendors: "best [category] software for [use case]", "how does [your product type] work", "[your category] for [industry]."

Have the team run each query across all four major AI platforms and score your brand's presence. The output is your baseline citation rate.

Step 3: Map competitor AI visibility gaps#

For each query in the set, your team should record which competitors appear and build a simple table: query on one axis, brand on the other, citation present or absent in each cell. Any query where a competitor appears and your brand does not is a prioritized content gap. As CMO, your job here is to rank those gaps by pipeline value, starting with the queries that sit closest to purchase intent.

Step 4: Assess content for AI citability#

Task your SEO manager or content team with scoring existing high-intent pages against the checklist below. Pages that fail three or more items should be prioritized for restructuring before new content is commissioned. Our free AEO Content Evaluator automates this scoring against the CITABLE framework.

Table 1: Technical audit checklist for AI citability

Step 5: Audit AI-sourced MQL sources#

Most CRMs have no field to capture AI-referred pipeline. Work with your marketing ops or RevOps team to put this in place before any optimization work begins. Here is the three-step attribution setup:

1. UTM parameters: Tag all links you control in platforms that support tracking. Use utm_source=chatgpt or utm_source=perplexity for any AI-platform referral traffic you can identify in Google Analytics 4 (GA4).
2. Self-reported attribution: Add a free-text "How did you hear about us?" field to every demo request and contact form. This single field captures the buyer research path that UTMs miss entirely.
3. CRM mapping: Create a custom field in HubSpot or Salesforce labeled "AI-referred source." Route the form field response into this field and tag deals sourced from AI platforms for quarterly pipeline review. This is not a perfect attribution model. Different data sources may show varying numbers for the same period. Acknowledge the discrepancy in board reporting and use a range rather than a single figure. It is more defensible than no measurement, and sufficient for quarterly board reporting when you present it as a range with stated confidence intervals.

Optimizing your budget for AI search audits#

Diagnosing visibility vs. performance gaps#

A visibility gap means competitors are cited in AI answers and you are not. A performance gap means you appear but it is not driving pipeline. These require different fixes.

Visibility gaps come from structural content issues (content is not extractable), entity issues (AI systems have weak associations between your brand and the category), or information consistency issues (your brand's claims are contradicted across sources). Performance gaps come from appearing in informational queries ("how does X work?") rather than commercial queries ("best X for Y use case"). Commercial citations correlate with pipeline because the buyer is in evaluation mode when they ask.

Run the Step 1-3 audit above to identify which problem you have before spending on fixes.

Reporting AI search ROI to the board#

AI-referred visitors tend to arrive pre-qualified because the AI assistant has already synthesized information about your product before they click through. Frame AI-referred pipeline to your board using the attribution path you built in Step 5: AI-referred sessions from GA4, MQL count from the CRM custom field, pipeline value from deals tagged as AI-sourced, and a conversion rate comparison against standard organic search.

Be transparent about measurement uncertainty. CMOs who present a range with stated caveats earn more credibility than those who present a single number that the CFO will immediately challenge. The measurement flaw we documented in AI tracking platforms is still relevant: most tools overstate precision because they use static prompt testing. A defensible model combines platform-level citation tracking, GA4 referral segmentation, and self-reported form data. Agreement across two of the three streams is sufficient to make a pipeline attribution claim with reasonable confidence.

Project roadmap for professional AI audits#

Defining audit scope and output#

A professional AI search audit should deliver five components:

1. Entity map showing how AI systems currently associate your brand with category, use cases, and competitors
2. Schema and structured data audit covering Organization, Product, FAQ, and HowTo markup
3. Citation gap analysis across ChatGPT, Claude, Perplexity, and Gemini for your priority query set
4. Content audit scoring existing high-intent pages against extraction criteria
5. Prioritized content roadmap with specific page targets and CITABLE framework specifications

Anything that delivers fewer than these five is a partial audit. See what a full-service managed AI visibility audit includes.

Expected timelines for AI visibility lift#

Set realistic expectations before committing budget:

• Weeks 1-2: Initial citations from new content begin appearing in AI responses
• Months 1-3: Citation rate on priority queries starts moving from baseline
• Months 3-4: Meaningful citation rate lift with consistent optimization
• Months 4-6: Continued optimization across all three surfaces with measurable pipeline attribution

These timelines reflect what we observe across our client base. Individual results depend on category competitiveness, content quality, and how consistently off-page consistency work is executed.

Flexible pricing for AI audit sprints#

Our Search Visibility Diagnostic is €4,370 one-off and includes CITABLE-optimized articles as a core output.

Table 2: AI audit pricing comparison

All retainers are month-to-month. Full deliverable breakdowns are on our pricing page.

DIY versus managed AI visibility audits#

DIY vs managed AI audit paths#

DIY audits are suitable for initial validation: confirming whether a gap exists, identifying the three or four highest-priority queries, and testing whether your existing content passes the extractability checklist above. Use this free AI visibility audit checklist to run Steps 1-5 above in one sitting. They require no budget beyond time and access to the major AI platforms.

Managed audits are necessary when you need systematic measurement across hundreds of queries, technical schema implementation, off-page consistency work, and a content production program running in parallel.

Staffing requirements for AI audits#

A complete AI search audit draws on three distinct skill sets. Before deciding between a DIY and managed path, assess whether your team covers all three:

• SEO specialist: Query mapping, technical schema implementation, and content architecture
• Content editor: Answer-first structure, section length discipline, and entity clarity
• AI/ML engineer: Dense passage retrieval mechanics and RAG system behavior

Most in-house marketing teams have the first two. Whether you have the third determines whether the audit recommendations you act on are grounded in how LLMs actually retrieve content, or in how someone assumed they would. Our citation tracking workflow guide shows what automated visibility measurement looks like when an engineering team builds the tooling rather than relying on third-party APIs alone.

Key considerations for your AI visibility audit#

Timeline for an AI visibility audit#

Our Search Visibility Diagnostic is a two-week engagement. It covers entity mapping, schema auditing, citation gap analysis across the major AI platforms, and production of CITABLE-optimized articles targeting priority buyer queries. That is enough to establish a baseline and identify the highest-impact gaps before committing to a longer engagement.

How citation rate grows in practice#

One anonymous B2B SaaS client (under NDA) grew AI-referred trials 6x in 7 weeks, reaching 3,500+ AI-referred trials total. The program focused on restructuring existing high-authority blog posts using the CITABLE framework, adding explicit entity relationships in copy, and building information consistency across Reddit, G2, and industry roundup articles. Initial citations appeared within two weeks of the first optimized page going live, with continued improvement through week 7.

The pattern holds across named clients as well. Before working with

Discovered Labs, incident.io's team had been experimenting with homegrown LLM prompts without a clear strategy for what to optimize or how to structure content for AI retrieval.

"I have recommended you to multiple peer CMOs. There are large organizations like Hubspot and Ramp who have dedicated teams to work on large projects like AEO. For everyone else (except my competitors) there's Discovered Labs!" - Tom Wentworth, CMO, incident.io

For Sova Assessment, organic search became the number-one pipeline channel, contributing more than 50% of year-to-date pipeline. For an anonymous B2B SaaS client (under NDA), AI-referred trials grew 6x in 7 weeks, reaching 3,500+ total, driven by citation rate improvements across ChatGPT, Claude, and Perplexity.

Does existing content require updates?#

No. An audit identifies which existing high-intent pages can be restructured using the CITABLE framework to perform across all three surfaces. In most cases, a meaningful share of existing blog content has the right topical coverage but the wrong structure for LLM passage extraction. Restructuring those pages rather than replacing them protects your existing ranking equity while adding citation potential. Our post on SEO and AEO differences explains what changes and what stays the same across the optimization process.

Managed AI audit pricing tiers#

Our three primary options:

• Search Visibility Diagnostic (€4,370 one-off): Two-week engagement. Entity map, schema audit, CITABLE-optimized articles, prioritized roadmap. No commitment beyond the one-off fee.
• Establish (€7,995/mo, or €6,995/mo on a 6-month commitment): Up to 20 SEO and AEO articles per month, visibility tracking, off-page consistency work, and a dedicated four-person pod.
• Compete (€12,995/mo, or €10,995/mo on a 6-month commitment): All Establish deliverables plus expanded content production, landing pages for high-intent keywords, syndication, and quarterly business reviews. All retainers are month-to-month. If we stop delivering, you leave. That is the accountability structure we prefer.

Conclusion#

An AI search audit is the step that makes optimization work. Without a baseline citation rate, a competitor gap map, and a content score against the CITABLE framework, optimization decisions are guesses. The three surfaces, web search, citations, and training data, each require different fixes, and the gap between them is where buyer attention currently goes unmeasured. Start with a DIY query set to confirm a gap exists, then decide whether a managed program is the right next step based on how many queries you need to cover and whether you have the engineering depth to act on the findings. If you want to start with a self-serve baseline, use our free AEO Content Evaluator to score existing pages against the CITABLE framework in minutes. If you want a full diagnostic with a team behind it, book a call and we will tell you honestly whether the Search Visibility Diagnostic is the right starting point for your situation.

FAQs#

What is an AI search audit?#

An AI search audit measures how often and how accurately your brand appears in responses from ChatGPT, Claude, Perplexity, and Gemini for your priority buyer-intent queries. It produces a citation rate baseline, competitor gap analysis, schema audit, and prioritized content roadmap covering web search, LLM passage retrieval, and training data.

How long does an AI visibility audit take?#

A professional managed audit typically takes four to six weeks to complete for comprehensive tracking and analysis. A DIY audit covering a small set of queries across four platforms can be completed more quickly, but it does not include schema implementation, off-page consistency analysis, or content production.

What is a good citation rate for B2B SaaS?#

Most programs that we work on start at low single-digit citation rates on priority queries before optimization. A well-optimized program can achieve significant citation rate improvements, as demonstrated by incident.io's progression from 38% to 64% AI visibility.

Do I need to hire an AI engineer to run an AI audit?#

No. A managed audit includes the AI/ML engineering expertise you need. DIY audits using tools like our AEO Content Evaluator require only SEO and content skills, though technical schema implementation will eventually require engineering support.

Can I track AI-referred pipeline in HubSpot or Salesforce?#

Yes, using UTM parameters on controlled links, a free-text "How did you hear about us?" field on forms, and a custom CRM field routing AI-referred sources into HubSpot or Salesforce. These three layers together produce a defensible attribution path for board reporting.

Does my existing blog content become useless for AI search?#

No. An audit typically finds that a significant share of existing high-intent content has the right topical coverage but wrong structure for LLM passage extraction. Restructuring those pages using the CITABLE framework protects your existing Google ranking equity while adding citation potential across AI platforms.

Key terms glossary#

AEO (Answer Engine Optimization): The practice of optimizing content to be cited by AI assistants and conversational search engines when answering user queries.

GEO (Generative Engine Optimization): Optimization techniques focused on appearing in responses from generative AI platforms that synthesize answers from multiple sources.

Citation rate: The percentage of priority buyer-intent queries where an AI platform mentions your brand in its response.

Dense passage retrieval (DPR): A dual-encoder architecture that maps questions and candidate passages into vector space to select semantically similar passages rather than ranking full documents.

CITABLE framework: Our content optimization methodology designed for LLM passage extraction: Clear entity and structure, Intent architecture, Third-party validation, Answer grounding, Block-structured for RAG, Latest and consistent, Entity graph and schema.

Share of voice: Your brand's citation count as a percentage of total citations in your category across a defined query set.

AI-referred MQL: A marketing-qualified lead where the first or last touch was an AI platform, tracked via UTM parameters and self-reported form fields.