QBiz Leads AI

AI Visibility Audit Example

This is what an AI visibility audit actually finds

Most agencies describe their audit process in vague terms: "we analyse your site," "we identify gaps," "we deliver recommendations." That tells you nothing about what you are paying for. Below is a complete walkthrough of an AI visibility audit on a realistic business, including the checks, findings, scores and recommended order of work.

The subject is a fictional plumbing company based in Manchester. The business, the website, the findings and the recommendations are all representative of what we encounter in real audits of service businesses across the UK. The specifics are invented. The patterns are not.

Get your own audit

What an audit is for

An AI visibility audit answers one question: are the technical and content signals clear enough for answer engines to interpret your business?

That question has technical answers: documented checks against known crawl, schema and content-readability signals, plus observed source-selection patterns. The audit below walks through each one.

01

It is not a website review

A website review looks at design, usability and conversion. An AI visibility audit looks at what happens when ChatGPT, Perplexity or Google AI Overviews try to read your site programmatically: whether crawlers can access it, whether schema explains it, and whether the page content can be extracted cleanly. A site can score well on every traditional metric and still be invisible to AI platforms because of structural issues that do not affect human visitors.

02

It produces a specific deliverable

The output is not a slideshow or a vague summary. It is a documented report: what was checked, what was found, what the severity is, what the fix is and what order to do the fixes in. Every finding is tied to a specific technical signal. Every recommendation has a reason and a priority. You can hand the report to a developer and they can act on it without further briefing.

03

It tells you where to spend first

Not every fix carries equal weight. A blocked crawl path that prevents AI platforms from reading your site at all is more urgent than a missing FAQ schema on a low-traffic page. The audit prioritises findings by commercial impact: which fixes will affect the most revenue-relevant pages, which address the most common buyer queries, which unblock the largest number of AI-platform interactions.

The audit subject

Meet the business: Clarkson & Sons Plumbing, Manchester

This is a fictional company built to represent a common audit subject: a well-established local service business with a professional website, decent Google rankings and no idea whether AI platforms can read any of it. The details below mirror what we typically receive at the start of an engagement.

What the audit checks

Every AI visibility audit examines six technical areas. Each one affects whether an AI platform can understand your business well enough to include it in a generated answer. Here is what each check involves and what we looked at on the Clarkson & Sons site.

1. Schema markup

Is structured data present? Is it correct? Does it cover the business entity, individual services, locations and FAQs? We check for LocalBusiness, Service, FAQPage and any other relevant schema types. We also check whether the schema validates without errors.

2. Crawl access

Can AI-specific user agents (GPTBot, Google-Extended, PerplexityBot, ClaudeBot) reach the site's pages? We review robots.txt, server-level rules, security plugins, CDN settings and firewall configurations. We also check for the presence and quality of an llms.txt file.

3. Metadata quality

Do title tags, meta descriptions and Open Graph data contain clear service definitions and location signals that AI platforms can read and use? Or are they written purely for human click-through with no machine-readable structure?

4. Page hierarchy

Does the site's structure communicate the relationship between services, locations and supporting content? A flat architecture with 28 disconnected pages tells an AI system nothing about scope or specialism.

5. FAQ coverage

Are FAQ sections present on service pages? Do the questions match what buyers actually ask AI platforms? Are the answers structured as self-contained, extractable blocks? Is FAQPage schema applied correctly?

6. Entity consistency

Does the business name, service terminology, location formatting and credential referencing appear the same way across every page? Inconsistencies confuse AI systems trying to confirm what the business actually is.

Audit findings

What the audit found on the Clarkson & Sons website

Below are the findings, organised by severity. Critical findings prevent AI platforms from reading the site at all. Warnings reduce accuracy or completeness. Informational findings highlight improvements that would strengthen AI readability but are not blocking issues. Passes confirm what is already working.

Critical: Crawl access blocked
Blocking AI platforms entirely

Wordfence firewall blocking AI user agents

The site uses Wordfence with its default "advanced blocking" rules enabled. These rules classify GPTBot, ClaudeBot and PerplexityBot as suspicious traffic and return 403 responses. ChatGPT, Claude and Perplexity cannot read any page on the site. The site owner was unaware of this setting.

Additionally, the robots.txt file contains a blanket Disallow: / directive for several AI-specific user agents, added by the web agency as a precautionary measure during the original build. This doubles the block: even if the firewall were fixed, the robots.txt would still instruct AI crawlers to ignore the entire site.

Critical: Schema markup absent
No structured data on any page

Zero schema markup across all 28 pages

The site has no structured data of any kind. No LocalBusiness schema identifying the company. No Service schema for any service offering. No FAQPage schema. No Organisation schema. When an AI platform crawls this site (assuming the access block is lifted), it has to infer everything from the visible text. That means the AI is guessing the business name, guessing which services are offered, guessing the service area and guessing the relationship between pages. Every guess is a potential inaccuracy in a generated answer.

Warning: FAQ sections missing or thin
Reducing extractable content

Only 2 of 8 service pages have FAQ sections

The boiler installation page has four FAQs. The emergency plumbing page has three. The remaining six service pages (bathroom fitting, drain clearance, leak detection, radiator installation, underfloor heating, commercial plumbing) have no FAQ content at all. The existing FAQs are also written from the business's perspective ("Why choose Clarkson & Sons?") rather than the buyer's ("How much does a boiler installation cost in Manchester?" or "How quickly can an emergency plumber arrive?").

FAQ sections are the most extractable content format for AI platforms. Each question-answer pair is a self-contained unit that maps directly to the queries buyers type into ChatGPT and Perplexity. Missing FAQs on service pages is a missed opportunity on every buyer query those pages should be answering.

Warning: Metadata written for humans only
Weak machine-readable signals

Title tags and meta descriptions lack service and location specificity

The boiler installation page's title tag reads: "Boiler Installation | Clarkson & Sons." No location. No service qualifier. No indication of residential vs. commercial. The meta description reads: "Looking for a new boiler? We can help. Call us today for a free quote." That communicates nothing an AI system can use to match the page to a specific buyer query about boiler installation in Manchester.

Across the site, 22 of 28 pages follow the same pattern: short, generic title tags and meta descriptions written for click-through rather than machine reading. AI platforms use metadata as a primary signal when deciding what a page covers. Vague metadata produces vague representation.

Warning: Page hierarchy is flat
Reducing contextual understanding

All 8 service pages sit at the same level with no grouping

The site has eight service pages, all linked directly from the main navigation with no parent-child relationship. There is no distinction between core services (emergency plumbing, boiler installation) and specialist services (underfloor heating, commercial plumbing). There is no hub page connecting residential services or grouping location-specific offerings. From an AI platform's perspective, the site appears to offer eight disconnected services with no indication of which is the primary offering, which serves which area, or how they relate to each other.

Informational: Entity inconsistencies
Minor signal confusion

Business name appears in three different formats

The homepage header reads "Clarkson & Sons Plumbing." The footer reads "Clarkson and Sons." The about page reads "Clarkson & Sons Plumbing Services Ltd." The Google Business Profile uses "Clarkson & Sons Plumbing Manchester." These variations are natural for a business that has evolved over time, but AI platforms treat each as a potentially different entity. Standardising to a single, consistent name across all pages and external profiles strengthens the signal.

Service terminology shows similar inconsistency. The boiler page alternates between "boiler installation," "boiler fitting," "new boiler installation" and "boiler replacement" without clarifying whether these are the same service or different offerings. An AI system reading the page cannot determine whether the business offers one boiler service or four.

Informational: No llms.txt file

No llms.txt file present

An llms.txt file provides AI platforms with a plain-text summary of the site's purpose, structure and key pages. It functions as a table of contents for large language models. The Clarkson & Sons site does not have one. For a site this size (28 pages), an llms.txt file would help AI platforms quickly identify the business, its services and its service areas without crawling every page individually.

Pass: Page load speed
Working correctly

All pages load under 2.5 seconds with clean HTML

The site is well-built from a performance standpoint. Pages load quickly, the HTML is clean, there are no render-blocking scripts interfering with crawl, and the server responds reliably. This is a solid foundation. The AI visibility issues are structural and content-related, not performance-related.

Pass: Google Business Profile
Strong external signal

GBP is optimised with 140+ reviews and accurate details

The Google Business Profile is well-maintained: correct address, phone number, service categories, opening hours and 140+ reviews with a 4.7 average. AI platforms (particularly Google AI Overviews) use GBP data as a trust signal. This is already working in the business's favour and provides a strong external entity reference that schema markup on the website should connect to.

The report

What the audit report looks like

Every audit produces a structured report. Below is a condensed version of what the Clarkson & Sons report contains. The real report includes detailed per-page breakdowns, before-and-after code examples and implementation notes for each recommendation. This preview shows the format and the level of specificity you can expect.

23/100 Overall AI readiness The site scores well for speed and basic HTML quality but fails on every AI-specific signal: no schema, blocked crawlers, thin FAQs, weak metadata.
2 Critical issues Crawl access and schema markup. Both must be resolved before any other optimisation work will have an effect.
5 Improvement areas FAQ coverage, metadata, page hierarchy, entity consistency and llms.txt. Each has a documented fix with priority ranking.
AI Visibility Audit Report , Clarkson & Sons Plumbing

Schema Markup

LocalBusiness schema Not present
Service schema (per service page) Not present (0 of 8)
FAQPage schema Not present
Organisation schema Not present

Crawl Access

GPTBot (ChatGPT) Blocked (firewall + robots.txt)
ClaudeBot (Claude/Anthropic) Blocked (firewall + robots.txt)
PerplexityBot Blocked (firewall + robots.txt)
Google-Extended Blocked (robots.txt only)
Googlebot (standard) Permitted
llms.txt Not present

Metadata Quality

Title tags with service + location 6 of 28 pages
Meta descriptions with entity markers 0 of 28 pages
Open Graph data Present but generic

FAQ Coverage

Service pages with FAQs 2 of 8
FAQs matching buyer queries 3 of 7 questions
Answer format (extractable) Partial

Entity Consistency

Business name variants found 4 variants
Service terminology consistency Inconsistent on 5 pages
Location formatting Consistent

Page Hierarchy

Service page grouping Flat (no parent-child)
Internal linking depth 1 level (nav only)
Breadcrumb navigation Not present

Recommendations

What we would recommend: prioritised by commercial impact

The audit does not just list problems. It produces a priority-ordered action plan. For Clarkson & Sons, the order is simple: access first, accuracy second, coverage third. Fix the crawl block before adding schema. Correct the business and service signals before expanding FAQs. Then widen the range of buyer questions the site can answer.

  1. 01

    Unblock AI crawl access (critical, week 1)

    Adjust Wordfence firewall settings to whitelist GPTBot, ClaudeBot, PerplexityBot and Google-Extended. Remove the blanket Disallow: / directives for AI user agents from robots.txt. Test each user agent to confirm 200 responses on all service pages. This is the single highest-priority fix because every other recommendation is irrelevant while AI platforms cannot read the site. Until crawl access is resolved, the site is invisible to every AI answer engine regardless of its content quality.

  2. 02

    Implement schema markup (critical, weeks 1-2)

    Add LocalBusiness schema to the homepage with accurate business name, address, phone, opening hours, service area and business type. Add Service schema to each of the eight service pages with service name, description, provider reference and area served. Add FAQPage schema to the two pages that already have FAQ sections. Add Organisation schema linking to the Google Business Profile. Validate all schema through Google's Rich Results Test and Schema Markup Validator. This gives AI platforms a structured, machine-readable description of the business instead of forcing them to guess from visible text.

  3. 03

    Rewrite metadata for machine reading (high priority, week 2)

    Update title tags on all 28 pages to include the specific service, location and business entity. Example: "Boiler Installation Manchester | Clarkson & Sons Plumbing" instead of "Boiler Installation | Clarkson & Sons." Update meta descriptions to include service definition, area served and a clear statement of what the page covers. Example: "Gas boiler installation for homes in Manchester, Salford and Stockport. Clarkson & Sons Plumbing: Gas Safe registered, same-week installation, free site survey." Update Open Graph data to match. This ensures that when an AI platform reads the page's metadata, it can immediately identify the service, the location and the provider without reading the full page content.

  4. 04

    Expand FAQ sections on all service pages (high priority, weeks 2-3)

    Add FAQ sections to the six service pages that currently lack them. Rewrite the existing seven FAQs on the boiler and emergency pages so the questions match what buyers actually ask AI platforms: "How much does a boiler installation cost in Manchester?", "How quickly can an emergency plumber get to me in Salford?", "What is included in a drain clearance service?" Each answer should be a self-contained paragraph: concise enough for an AI system to extract directly, specific enough to be genuinely useful, and structured with FAQPage schema so the question-answer pair is individually addressable. Target: 5-8 FAQs per service page, covering cost, process, timeline, qualifications, area coverage and common buyer concerns.

  5. 05

    Restructure page hierarchy and standardise entities (medium priority, weeks 3-4)

    Group service pages under logical parent categories: "Residential Plumbing" (emergency, bathroom fitting, leak detection) and "Heating & Installation" (boiler installation, radiator installation, underfloor heating) with "Commercial Plumbing" as a separate branch. Add breadcrumb navigation with BreadcrumbList schema. Create internal links between related services so AI platforms can follow the relationship between offerings. Standardise the business name to "Clarkson & Sons Plumbing" across all pages, the footer, the about page and external profiles. Standardise service terminology so each service page uses one consistent term for its offering. Add an llms.txt file summarising the business, its services, its service areas and its key pages. These changes strengthen the contextual signals AI platforms use when deciding how comprehensively to represent the business.

After the audit

What would happen if Clarkson & Sons implemented these recommendations

An audit is a diagnostic tool, not a guarantee. But the changes it recommends have specific, predictable effects on how AI platforms interact with a website. Here is what would change for Clarkson & Sons if the recommendations above were implemented.

AI platforms can read the site

The most immediate change. ChatGPT, Claude, Perplexity and Google AI Overviews go from being unable to access the site at all to having full crawl access to every page. This is the prerequisite for everything else. No amount of content quality matters if AI platforms receive a 403 error when they try to read it.

AI platforms know what the business does

Schema markup gives AI systems a structured, machine-readable answer to "what is this business, what does it do, and where does it operate?" Instead of guessing from paragraph text, AI platforms can read and use the business entity, its services, its locations and its FAQs programmatically. This reduces the chance of inaccurate representation in generated answers.

The site answers more buyer queries

Expanded FAQ sections, rewritten metadata and a clearer page hierarchy mean the site can be matched to a wider range of buyer queries. When someone asks ChatGPT "how much does a boiler installation cost in Manchester?" the site now has a specific, extractable answer on a page that AI platforms can both access and understand. Before the audit, neither the content nor the access existed.

Common questions

Questions about AI visibility audits

Is the example above based on a real business?

No. Clarkson & Sons Plumbing is fictional. The business, the website, the findings and the recommendations were constructed to illustrate what an AI visibility audit typically produces. The patterns are real: crawl-access blocks from security plugins, missing schema markup, thin FAQ sections, vague metadata and entity inconsistencies are the most common issues we find on service-business websites. The specific details are invented to show how those patterns appear in practice.

How long does an AI visibility audit take?

A typical audit for a site with 10-50 pages takes 3-5 working days from start to delivered report. Larger sites or sites with complex technical configurations may take longer. The timeline includes the crawl-access check, schema review, metadata analysis, page-hierarchy mapping, FAQ assessment, entity-consistency review and the production of the prioritised recommendation report.

What do I receive at the end?

A structured report covering every check area: schema markup, crawl access, metadata quality, page hierarchy, FAQ coverage and entity consistency. Each finding includes the severity level, a description of the issue, the specific pages affected and a recommended fix. The report also includes an overall AI readiness score and a priority-ordered action plan that you can hand directly to a developer or use as a brief for implementation work.

Do you also implement the recommendations?

Yes. QBiz Leads offers both standalone audits and audit-plus-implementation packages. The audit tells you what needs fixing. The implementation work (covered under our AI optimisation services) carries out the technical changes: schema implementation, crawl-access fixes, metadata rewrites, FAQ expansion, page-hierarchy restructuring and entity standardisation. You can also take the report to your existing developer or agency if you prefer to handle implementation internally.

How is this different from an SEO audit?

An SEO audit focuses on traditional search signals: keyword targeting, backlink profile, page speed, Core Web Vitals, index status and ranking positions. An AI visibility audit focuses on AI-specific signals: whether AI crawlers can access your site, whether structured data is present and correct, whether your content is formatted for extraction by answer engines, and whether your entity signals are consistent enough for AI platforms to confidently represent your business. The two audits examine different layers of the same website. A site can pass an SEO audit and fail an AI visibility audit because the skills are different.

What if my site already has some schema markup?

The audit checks whether existing schema is correct, complete and validated. Partial schema can be worse than no schema if it contains errors, uses the wrong types or provides inaccurate information. We frequently find sites with schema that was auto-generated by a plugin but never reviewed: wrong business type, missing service entries, incomplete location data or FAQPage schema applied to content that is not actually in FAQ format. The audit identifies what is correct, what needs fixing and what is missing.

Can I see a sample of the full report before commissioning an audit?

The walkthrough on this page is a condensed version of what a full report contains. The actual report includes per-page breakdowns, code-level examples (before-and-after schema, metadata rewrites), crawl-test results for each AI user agent, and detailed implementation notes for each recommendation. If you would like to see the report format before commissioning, request a free AI visibility check: it covers the headline findings and gives you a clear sense of the depth and format.

How often should an AI visibility audit be repeated?

AI platforms update their models, adjust how they handle sources and change which signals appear to matter. A site that passes an audit today may develop new issues after a CMS update, a plugin change, a redesign or a shift in how AI platforms process structured data. A follow-up audit every 6-12 months is usually sensible, or immediately after any significant change to the site's technical stack, design or content structure.

AI Visibility Services AI Optimisation Services ChatGPT SEO Answer Engine Optimisation Google AI Overview SEO AI SEO for Local Businesses

Find out what an AI visibility audit would uncover on your site

The example above shows the depth and specificity of the process. Your site has its own technical profile, structural gaps and priority fixes. A free AI visibility check gives you the headline findings: which AI platforms can access your site, whether your schema markup exists and validates, and where the largest gaps sit. From there, you can decide whether a full audit is worth commissioning.

Request your free check