Somewhere today, a procurement head in Pune typed a question into ChatGPT that your sales team would love to answer. The reply named three vendors. It quoted a pricing page, a Reddit thread and a listicle from two years ago. Your brand did not make the cut, and nobody on your team will ever know that conversation happened.
That invisible shortlist is the reason an AI search visibility audit exists. Rankings still matter, but buyers now meet brands inside answers first, and those answers rarely send a click. This guide walks through the seven checks iTechSEO runs for B2B firms, D2C brands and local service businesses across India, the UAE and the UK. Every check ends in the same place: a lead in your CRM, or a clear reason to stop spending.
Key takeaways
|
What Is an AI Search Visibility Audit?
An AI search visibility audit is a structured review of how often AI answer engines such as Google AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini and Microsoft Copilot mention, cite and link your brand for the questions buyers ask, how accurately they describe you, and how many of those appearances turn into leads and revenue.
It differs from a rankings report in one important way. A rankings report asks where you sit on a list. An AI visibility audit asks two harder questions: did you make the answer at all, and did the answer tell the truth about you?
A complete audit tracks eight numbers.
| Metric | What it tells you |
| Mention rate | Share of tested prompts where an engine names your brand |
| Citation rate | Share of prompts where an engine links or footnotes one of your URLs |
| Share of voice | Your brand mentions divided by all brand mentions across the same prompt set |
| Accuracy score | How closely services, locations, pricing and positioning match reality |
| Sentiment | If the framing helps or hurts your place on a shortlist |
| AI impressions | Google AI Overview and AI Mode impressions from Search Console |
| Copilot citations | Citations and grounding queries from Bing Webmaster Tools |
| AI referred leads | Sessions from AI referrers that reach your CRM as leads or deals |
Why Rankings Alone Now Miss Half the Story
Buyers already shop this way. Forrester’s State of Business Buying 2024 found that almost 95% of B2B buyers expected to use generative AI in their purchase process within the following 12 months.
Clicks behave differently too. Pew Research Center tracked the browsing of 900 US adults and found that people clicked a traditional result on 8% of visits when Google showed an AI summary, against 15% when it did not. Only 1% clicked a link inside the summary itself (July 2025).
Yet a citation still pays. Seer Interactive studied 3,119 terms in November 2025 and found organic CTR of 0.70% on AI Overview queries where the brand was cited, against 0.52% where it was not. The traffic that does arrive tends to buy. Adobe Digital Insights reported in April 2026 that AI referred visits to US retail sites converted 42% better than other traffic in March 2026, reversing a 38% deficit from a year earlier.
The measurement gap has narrowed as well. Google rolled Generative AI performance reports out to Search Console worldwide on 31 August 2026, although sites with few AI impressions may not see data yet. Bing Webmaster Tools has reported Copilot citations since a February 2026 preview. For the first time you can audit AI visibility with first party data instead of guesswork. For the strategy side of the same shift, see our full AI search optimization playbook.
Before You Start: Set Up an Audit You Can Repeat
AI answers change from one run to the next. Ask ChatGPT the same question three times and you may see three different shortlists. An audit that ignores this produces screenshots, not evidence. Set four ground rules before you test anything.
- Fix your scope. List the engines (Google AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini, Copilot), the markets and languages you sell in, and every brand entity worth tracking: company name, product names, founders and common misspellings.
- Pick three to five real competitors. Use the names your sales team loses deals to, not the brands you admire.
- Control the test. Use a logged out or fresh session, set your location to the market you sell in, and run each prompt three times. Count a brand as present only when it appears in at least two of the three runs.
- Log everything in one sheet. Record the prompt, buyer stage, engine, date, run number, brands named, URLs cited, your position in the answer, accuracy notes and sentiment.
A pulse check on ten prompts fits inside an hour. The full audit takes longer because it touches analytics, the CRM and content owners, so book time with each of them before you begin.
If you have not yet decided which AI channels deserve budget, start with iTechSEO’s GEO, AEO and agentic SEO growth strategy, then come back with a shortlist of engines to audit.
The Seven Checks at a Glance
| # | Check | The question it answers | Where the evidence lives |
| 1 | Buyer question map | Which questions decide deals, and who do AI tools name for them? | Sales notes, proposals, support tickets, prompt log |
| 2 | Crawl, index and AI crawler access | Can every engine reach and quote the pages you want cited? | Search Console, robots.txt, CDN and firewall logs |
| 3 | Entity and answer accuracy | Do AI tools describe your company correctly? | Prompt log, Google Business Profile, directories, schema |
| 4 | Revenue first attribution | Which AI appearances produce leads and deals? | Search Console, Bing Webmaster Tools, GA4, CRM |
| 5 | Brand mentions and share of voice | How often do trusted sources name you compared with rivals? | Prompt log, mention tools, cited source lists |
| 6 | Conversion readiness | Do AI referred visitors find the answer and one next step? | GA4 landing pages, page speed data |
| 7 | Freshness and consolidation | Are the pages AI tools cite current and unique? | Content inventory, citation dates, change log |
Check 1: Map the Buyer Questions That Decide Deals
Start with 30 to 50 questions that buyers ask before they pick a vendor. Pull them from sales call notes, lost deal reasons, proposal Q&A and support tickets, not from a keyword tool alone. Keyword tools show what people type into Google. Your sales notes show what they ask when money is on the table.
Tag every question by buyer stage:
- Problem: “Why are our demo requests falling while traffic grows?”
- Category: “What should a technical SEO review cover in 2026?”
- Comparison: “In house SEO team or agency for a Series A SaaS company in Bengaluru?”
- Vendor choice: “Which SEO agencies in India report revenue instead of rankings?”
- Branded: “Is iTechSEO a good fit for B2B lead generation?”
Run each question through each engine under the ground rules above, then record which brands and sources appear.
| Warning Do not spin out hundreds of near identical variants to inflate coverage. Google’s guide to optimizing for generative AI features calls pages built for every query variation unhelpful and warns they can breach the scaled content abuse policy. |
What good looks like: a ranked list of questions, each tagged with stage and deal value, plus the brands AI tools currently name for it.
Check 2: Confirm Crawl, Index and AI Crawler Access
An engine cannot cite a page it cannot reach. This check sounds basic, yet it often hides the most expensive mistake in the whole audit: a firewall rule nobody remembers adding.
Start with Google. Its AI features documentation says a page must be indexed and eligible to show with a snippet before it can appear as a supporting link in AI Overviews or AI Mode. Look for stray noindex tags, nosnippet or tight max-snippet directives, and canonical tags that point somewhere else.
Next, check the new switch. Since 31 August 2026, Search Console includes a generative AI control that lets a site opt out of grounding AI Overviews, AI Mode and AI Overviews in Discover. Google says the setting is not a ranking signal for regular web results. Confirm nobody changed it by accident, and document the decision if leadership chose to opt out.
Then review every crawler that feeds an answer engine you care about.
| User agent | Owner | What it controls | Audit action |
| Googlebot | Search results, AI Overviews and AI Mode | Must be allowed | |
| Google-Extended | Gemini model training and Gemini app grounding. It does not affect AI Overviews or AI Mode | Decide by policy | |
| OAI-SearchBot | OpenAI | Appearance in ChatGPT search answers | Allow for ChatGPT citations |
| GPTBot | OpenAI | Training of OpenAI foundation models only | Decide by policy |
| ChatGPT-User | OpenAI | Pages fetched when a user asks ChatGPT to visit them. robots.txt may not apply | Check firewall rules |
| Bingbot | Microsoft | Bing results and Copilot answers | Must be allowed |
| PerplexityBot | Perplexity | The Perplexity search index | Allow for Perplexity citations |
| Claude-SearchBot | Anthropic | Search results inside Claude | Allow for Claude citations |
OpenAI’s crawler documentation is clear on the difference that trips most teams up: sites that block OAI-SearchBot will not appear in ChatGPT search answers, while GPTBot only governs training. You can block one and allow the other.
robots.txt is only half the story. CDN bot protection and web application firewalls often turn AI crawlers away before they ever read robots.txt. Ask your hosting or security team for 30 days of server logs filtered by these user agents and look for 403 responses.
Finally, check speed and rendering. Google’s Core Web Vitals guidance sets good thresholds at LCP within 2.5 seconds, INP at 200 milliseconds or less and CLS at 0.1 or less, measured at the 75th percentile. Several AI crawlers read raw HTML without running JavaScript, so make sure your key answers sit in the server rendered page. For the wider technical list, work through these nine common technical SEO issues.
AI agents that compare vendors and book calls for buyers need the same open doors. Our guide on how AI search agents read and act on your pages explains what they look for once they arrive.
What good looks like: every priority page indexed, snippet eligible, reachable by the crawlers you chose, and fast on mobile.
Check 3: Test Entity and Answer Accuracy
Ask ChatGPT, Gemini, Perplexity, Copilot and Google AI Mode the same five prompts: “What does [brand] do?”, “Where is [brand] based?”, “Who does [brand] work with?”, “How much does [brand] cost?” and “What are the alternatives to [brand]?”
Score each answer on a four level rubric:
- Accurate: matches your current services, locations and positioning.
- Partly accurate: right category, wrong detail, such as an old address or a retired service.
- Wrong: confuses you with another company or invents facts.
- Missing: the engine does not know you.
Then trace every error to its source. AI tools usually repeat something they read: an outdated directory listing, an old press release, a LinkedIn page nobody updated, or your own blog post from 2021. Fix the source, not the symptom. Update your Google Business Profile, directory listings and About page first, because engines lean on them most for basic facts.
Keep markup honest. Google’s guide says you do not need new machine readable files, AI text files, special markup or llms.txt to appear in its generative features, and that structured data is not required. Keep Organization, Service and Article markup anyway, because it supports rich results and removes ambiguity, but make sure it matches what the page actually says. Place a direct two or three sentence answer under each key heading. Google also says there is no need to chop content into tiny pieces, so write for people and let clear structure do the work.
What good looks like: at least four of five engines describe you accurately, and every error has an owner and a fix date.
Check 4: Tie AI Visibility to Revenue
This is the check most audits skip and the one your finance team cares about. Treat every AI metric as a leading indicator until it produces an attributed lead. Four data sources form one chain.
- Search Console Generative AI report. It shows the impressions your pages earned in AI Overviews and AI Mode by page, country, device and date. It shows no clicks, CTR or queries, so use it for trends and page level baselines. Clicks from AI Overviews still land in GA4 as Google organic.
- Bing Webmaster Tools AI Performance. It shows citation counts, cited pages and the grounding queries behind them across Copilot and Bing. A June 2026 update added intents, topics, citation share and period comparison. It is one of the few free sources of query level data for AI answers.
- GA4. Create a custom channel group called AI Assistants and match session source against a regex such as chatgpt\.com|perplexity\.ai|gemini\.google\.com|copilot\.microsoft\.com|claude\.ai. OpenAI adds utm_source=chatgpt.com to ChatGPT referral links, which makes that traffic easy to isolate.
- CRM. Pass the source into hidden form fields and carry it onto the lead and deal record. Import offline conversions when deals close over the phone.
Track only what consent allows. India’s Digital Personal Data Protection Rules 2025 take full effect on 13 May 2027, the UAE applies Federal Decree-Law No. 45 of 2021, and the UK applies PECR alongside UK GDPR.
What good looks like: a monthly report where AI impressions and citations sit under leads and pipeline from AI referrers, never above them.
Check 5: Measure Brand Mentions and Share of Voice
AI tools recommend brands that other credible sources already discuss. Ahrefs found in May 2025 that branded web mentions correlated 0.664 with AI Overview visibility, against 0.218 for backlinks.
Turn your prompt log into one number with a simple formula:
| AI share of voice = your brand mentions ÷ all brand mentions across the prompt set × 100 Example: across 40 prompts, three runs each, the engines named brands 300 times in total and yours 36 times. Your AI share of voice is 12%. |
Calculate it per engine, because ChatGPT, Perplexity and Copilot draw on different sources. Bing’s citation share metric gives you the same view for Copilot citations without any manual counting.
Next, run the gap analysis. For every prompt where a competitor appears and you do not, open the cited sources. You will usually find a short list of roundups, analyst lists, podcasts, Reddit threads, review sites and trade bodies. That list becomes your outreach plan. Also check existing mentions for accuracy, since a wrong fact on a trusted site travels fast.
Grow mentions the durable way. Google’s spam policies prohibit buying links for ranking, and its AI guide calls the pursuit of inauthentic mentions unhelpful. Original data, expert commentary, real partnerships and useful free tools earn mentions that last.
What good looks like: a share of voice baseline per engine, plus a ranked list of third party sources to earn.
Check 6: Make AI Referred Pages Convert
Clicks from AI answers are rare, so each one carries more weight. Pull the landing pages that receive AI referrals from GA4, add the pages Search Console and Bing show as cited, and test each one against five questions:
- Does the answer appear on the first screen, in plain words?
- Does proof sit next to the claim: client names, certifications, partner badges, real numbers?
- Does the page load fast on a mid range Android phone over mobile data?
- Is there one next step that fits the stage, such as a demo for vendor choice or a guide for problem stage visitors?
- Does the page type match the question? Comparison questions need comparison pages, not the homepage. Pricing questions need a page that talks about pricing.
What good looks like: every AI referred landing page answers fast, proves it, and asks for one thing.
Check 7: Refresh, Merge or Retire
AI assistants lean toward fresher sources. Ahrefs analysed about 17 million citations across seven platforms in July 2025 and found the average URL cited by AI assistants was 1,064 days old, against 1,432 days for Google organic results, which makes AI citations 25.7% fresher. ChatGPT showed the strongest pull toward recent pages, while Google AI Overviews cited content slightly older than organic results.
Changing a date without changing the content fixes nothing. Look for three patterns and treat each one differently:
- Overlapping posts that split authority on one topic: merge them into the stronger URL.
- Outdated statistics, product names or regulations: update them and note the review date.
- Thin pages built for query variations: redirect or remove them, since they drift toward scaled content abuse.
Check that your own posts agree with each other as well. Conflicting advice across one site hands AI tools two versions of the truth to quote. Assign an owner and a review date to every commercial page, and log each change so you can match updates to movement in citations and leads.
What good looks like: every commercial page has an owner, a review date and a change log entry from the last 90 days.
Score Your Audit: The Seven Check Scorecard
Turn your findings into one number the leadership team can track. Score each check 0, 1 or 2, add them up, and rescore every quarter. Copy the table into a sheet and it becomes the front page of your AI search visibility audit report.
| Check | 0 points | 1 point | 2 points |
| 1. Buyer questions | No prompt list | List exists but has no stages or deal value | 30 or more questions tagged by stage and deal value |
| 2. Access | Priority pages blocked or not indexed | Minor gaps, such as a firewall block on one engine | All priority pages indexed, snippet eligible and reachable |
| 3. Accuracy | Most engines get core facts wrong | Two or three engines accurate | Four or more engines accurate |
| 4. Attribution | No AI channel in GA4 | GA4 channel but no CRM link | AI source carried through to deals |
| 5. Share of voice | No baseline | Baseline on one engine | Baseline on every priority engine plus a source gap list |
| 6. Conversion | Cited pages lack answers or a CTA | Some pages fixed | Every AI referred page answers fast with one CTA |
| 7. Freshness | No review dates | Review dates on some pages | Owner, date and change log on every commercial page |
- 0 to 5: fix access and accuracy before you spend on content or outreach.
- 6 to 10: the base works. Focus on attribution and share of voice next.
- 11 to 14: strong position. Move to monthly optimisation and test new engines.
Tools That Speed Up the Audit
| Tool | What it adds | Cost |
| Google Search Console (Generative AI report) | AI Overview and AI Mode impressions by page, country, device and date | Free |
| Bing Webmaster Tools (AI Performance) | Copilot citations, grounding queries, intents and citation share | Free |
| Google Analytics 4 | AI referrer sessions and conversions through a custom channel group | Free |
| Rich Results Test and Schema Markup Validator | Confirms structured data is valid and matches visible content | Free |
| Ahrefs Brand Radar | Brand mentions, cited pages and AI share of voice across engines | Paid |
| Semrush AI Visibility Toolkit | AI visibility score, mentions and competitor benchmarks | Paid, with a free checker |
| SE Ranking AI Visibility Tracker | Mentions and links across AI Overviews, AI Mode, ChatGPT, Gemini and Perplexity | Paid, with a free check |
Treat third party tools as directional. They sample prompts, and AI answers vary by session and location. Google’s guide also warns against tools that claim access to internal Google metrics. Your prompt log, Search Console, Bing Webmaster Tools and your CRM remain the source of truth for every AI search visibility audit.
Traditional SEO Audit vs AI Search Visibility Audit
| Area | Traditional SEO audit | AI search visibility audit |
| Main question | Where do we rank? | Do AI answers name, cite and describe us correctly? |
| Core data | Search Console Performance report, rank trackers | Search Console Generative AI report, Bing AI Performance, prompt log, GA4 AI channel |
| Crawl scope | Googlebot | Googlebot, Bingbot, OAI-SearchBot, PerplexityBot and other AI crawlers |
| Authority signal | Backlinks and referring domains | Brand mentions, linked and unlinked, plus backlinks |
| Success metric | Rankings and sessions | Citations and mentions that produce leads and pipeline |
| Review cadence | Once a year or after a traffic drop | Full audit each quarter, prompt checks monthly |
Five Mistakes That Waste an Audit
- Testing each prompt once. A single run is an anecdote. Three runs per engine show you a pattern.
- Counting mentions without reading them. A mention that calls you expensive or outdated can cost you the shortlist.
- Chasing files instead of fundamentals. llms.txt and AI specific markup will not fix a blocked crawler or a wrong address.
- Reporting impressions as results. Search Console AI impressions show reach, not revenue. Label them as leading indicators.
- Auditing only Google. Copilot, ChatGPT and Perplexity each draw on different indexes and sources, so each needs its own baseline.
Where iTechSEO Fits in Your AI Search Visibility Audit
iTechSEO has worked in search since 2010 and runs this AI search visibility audit for B2B firms, D2C brands and local service businesses across India, the UAE, the UK and global markets, from offices in Indore and Surat. The work covers buyer question mapping, crawler and index checks, entity fixes, AI attribution in GA4 and your CRM, and an action list ranked by revenue impact.
We work within Google Search Essentials, report honestly and never promise rankings or citations, because no one can. You keep every finding, prompt log and scorecard, and you can act on them yourself or with our team.
Frequently Asked Questions
What is an AI search visibility audit?
An AI search visibility audit is a structured review of how Google AI Overviews, AI Mode, ChatGPT, Perplexity, Gemini and Copilot mention, cite and describe your brand for buyer questions. It also checks crawler access, fact accuracy and attribution, so every appearance connects to leads.
How do I check if ChatGPT mentions my brand?
Open a logged out ChatGPT session, ask the questions buyers raise before they choose a vendor, and run each one three times. Record the brands and sources it names. Repeat monthly across Perplexity, Gemini, Copilot and Google AI Mode for comparison.
Can I see AI Overview traffic in Google Search Console?
Partly. The Generative AI performance report, live worldwide since 31 August 2026, shows AI Overview and AI Mode impressions by page, country, device and date. It shows no clicks or queries, so pair it with GA4 and CRM data.
How do I track Microsoft Copilot citations?
Use the AI Performance report in Bing Webmaster Tools. It shows how often Copilot and Bing AI answers cite your pages, which pages they cite, the grounding queries behind them and, since June 2026, your citation share for each query.
Do I need llms.txt or special schema to show up in AI Overviews?
No. Google’s guide to optimizing for generative AI features says you need no special files, AI text files or markup. Your page must be indexed and eligible for a snippet. Keep standard structured data for rich results and clarity.
Will blocking GPTBot or Google-Extended remove me from AI answers?
No. GPTBot governs OpenAI model training, while OAI-SearchBot controls ChatGPT search visibility. Google-Extended does not affect AI Overviews or AI Mode. The generative AI control in Search Console is the setting that removes you from Google’s AI features.
How long does an AI search visibility audit take?
A pulse check on ten buyer prompts takes under an hour. A full audit takes longer because analytics, CRM and content owners all contribute, so plan it as a short project with a clear owner rather than a single afternoon.
How often should I run an AI visibility audit?
Run a full audit every quarter and a prompt pulse check every month. Rerun it after a site migration, a rebrand, a major Google update or any change to your Search Console AI setting.
Start With the Ten Questions That Pay
Pick the ten buyer questions closest to a signed contract. Run each one three times in ChatGPT, Perplexity, Gemini, Copilot and Google AI Mode this week, and log who gets named. By Friday you will know exactly where your AI search visibility audit should begin. If you want a second pair of eyes on the results, book a review with iTechSEO.
| About the author Aziz Shaikh is Digital Marketing Head at iTechSEO, where he leads technical and AI search audits for clients across India, the UAE and the UK. https://www.linkedin.com/in/azizqshaikh/ |
You must be logged in to post a comment.