Search visibility is no longer only about where your website ranks on Google. A potential customer might ask Google an AI-powered question. They might ask ChatGPT to recommend a provider. They might use Microsoft Copilot to compare local companies. Your business could be mentioned. It could be linked as a source. It could be described incorrectly. A competitor might be recommended instead.
Or your brand might not appear at all. The difficult bit is measuring this properly. Searching your company once in ChatGPT and taking a screenshot is not an AI search visibility audit. A useful audit needs the same prompts, the same measurement rules and repeated tests over time. At Wixdek, we would measure six things:
- Does the brand appear?
- Is the information accurate?
- Is the business recommended when relevant?
- Is the website cited or linked?
- Does AI search visibility send referral visits?
- Do those visits produce leads?
This guide shows how to build that audit across Google, ChatGPT and Microsoft Copilot without pretending these platforms work in exactly the same way.
What Is an AI Search Visibility Audit?
An AI search visibility audit is a structured review of how a business appears across AI-powered search and answer systems. Instead of only checking traditional keyword positions, you test real customer questions and record what each platform says.
For example, a London business might want to know whether it appears when somebody asks:
- “Which SEO agencies in London work with service businesses?”
- That is very different from simply checking whether the business ranks for:
- SEO agency London
The AI audit records whether the company is mentioned, how it is described, whether it is recommended and which sources are used. You then repeat the same test later. That gives you a trend rather than one interesting screenshot.
AI Search Visibility Is Not One Metric
There is no universal AI search visibility number shared by Google, OpenAI and Microsoft. Each platform provides different information. Google began rolling out dedicated generative AI performance reporting in Search Console in June 2026. It can show impressions, pages, countries, devices and changes over time for supported generative AI Search features such as AI Overviews and AI Mode. Google initially made the report available to a subset of websites.
Microsoft introduced AI Performance in Bing Webmaster Tools in February 2026. The public-preview report shows when website content is cited across Microsoft Copilot, AI-generated Bing summaries and selected partner experiences. It can also show cited URLs and changes in citation activity over time. OpenAI allows publishers to track referral visits from ChatGPT using analytics, and ChatGPT adds utm_source=chatgpt.com to referral URLs. OpenAI does not provide website owners with the same type of Search Console-style visibility report described above.
So your audit needs to combine platform data with controlled manual testing.
What Should an AI Search Visibility Audit Measure?
Start with the same six areas for every platform.
1. Brand Appearance
First ask:
- Does the business appear at all?
- Imagine the prompt is:
- “Recommend SEO agencies for a London plumbing company.”
Record whether Wixdek appears. Do not only record the searches where it does. Missing mentions are useful data too.
2. Brand Accuracy
Being mentioned is not automatically good. Check whether the answer gets important facts right.
For a service business, that may include:
- business name,
- services,
- location,
- service areas,
- website,
- specialisms,
and other important public facts. A business mentioned with the wrong location or wrong service is not strong visibility. It is inaccurate visibility.
3. Recommendations
Next record whether the platform actually recommends the business when the prompt asks for providers.
There is a difference between:
- “Wixdek is an SEO agency.”
- and:
- “Wixdek is one option to consider for this requirement.”
Record the distinction. Do not treat every brand mention as a recommendation.
4. Citations and Links
Check whether your website is being used as a source.
Record:
- whether the site is cited,
- which URL is cited,
- what claim it supports,
and whether third-party sources are cited instead. This is particularly useful because a company can be discussed using information from another website rather than its own site.
5. Referral Visits
A citation becomes more commercially interesting when people actually visit. Track visits from AI platforms where reliable referral information is available. OpenAI says ChatGPT referral URLs automatically include utm_source=chatgpt.com, which can help publishers identify inbound traffic from ChatGPT search. Do not confuse citations with visits. A page can be cited many times without receiving many clicks.
6. Leads
Finally, ask what happened after the visit.
Did the visitor:
- call,
- submit a form,
- request a quote,
- book,
- buy,
- or take another meaningful action?
This is where AI search measurement moves beyond another visibility report.
Step 1: Decide Which Platforms You Will Audit
For this framework, use:
- Google Search and its generative AI experiences
- ChatGPT Search
- Microsoft Copilot and Bing’s AI search experiences
Do not assume that visibility on one means visibility on the others. They are separate systems.
Google’s generative Search experiences are connected with its existing Search systems and Search index. Google says normal SEO remains relevant because its generative AI features use core Search ranking and quality systems. Where available, Google’s dedicated generative AI Search Console report should form part of the audit.
ChatGPT
ChatGPT can search the web and provide links to relevant web sources. OpenAI says there is no guaranteed top placement. For content to be included in ChatGPT search summaries and snippets, publishers should allow OAI-SearchBot to access the relevant content. Manual prompt testing is still useful because publishers need to see how their business is actually represented to users.
Microsoft Copilot
Microsoft says web-grounded Copilot responses can use web search results and include links to the web pages used to support the answer. Bing Webmaster Tools now gives site owners an additional measurement source through its AI Performance report, which tracks citations across Microsoft AI experiences.
Step 2: Check Technical Access Before Testing Visibility
Do not start by rewriting content. First make sure the platforms can access what they need.
Check Google Access
For Google’s generative Search features, normal Google Search technical requirements still apply. Google says pages need to be indexed and eligible to appear in Search with a snippet to be eligible as supporting links in AI Search experiences. There is no special Google AI crawler requirement you need to add on top of normal Search SEO. Google also says you do not need special AI markup, llms.txt files or AI-specific schema to appear in its generative Search features.
Check ChatGPT Access
For ChatGPT Search, check OAI-SearchBot. OpenAI says publishers should make sure they are not blocking OAI-SearchBot if they want content to be discoverable, surfaced and cited in ChatGPT search. Also check server, firewall and CDN rules if the crawler still cannot reach the site.
OAI-SearchBot and GPTBot Are Not the Same
This distinction matters. OAI-SearchBot relates to ChatGPT search discovery. GPTBot is a separate control related to whether content may be used for model training. Blocking GPTBot does not mean the same thing as blocking OAI-SearchBot. OpenAI documents these controls separately. Do not put them into one generic “AI crawler” bucket.
Check Bing and Microsoft Access
Check that important public pages can be indexed through Bing. Bing Webmaster Tools should also be set up so you can access the AI Performance information available for your site. Microsoft’s AI Performance report can then help you see which pages are being cited across supported Microsoft AI experiences.
Step 3: Build a Fixed Prompt Set
This is the heart of the audit. Do not type whatever comes into your head each month. Create a fixed list. Keep the wording. Keep each prompt ID. Repeat it during the next audit. That gives you something you can compare.
Wixdek Fixed AI Search Visibility Prompt Set
The following 18-prompt structure can be adapted to most service businesses.
Branded Prompts
B01: What is [Brand]? B02: What services does [Brand] provide? B03: Is [Brand] a good option for [main service]? These test whether the system understands the business itself.
Service Discovery Prompts
S01: Which companies provide [service] in [location]? S02: Recommend companies for [service] in [location]. S03: Who should I consider for [service] in [location]? These test non-branded discovery.
Problem Prompts
P01: I have [customer problem]. What type of company should I contact? P02: How should I solve [customer problem]? P03: When should I hire a professional for [customer problem]? These test earlier stages of the journey.
Comparison Prompts
C01: What should I compare when choosing a [service provider]? C02: Which [service providers] in [location] should I compare? C03: What are good alternatives to [known competitor or provider]? These test commercial investigation.
Cost and Decision Prompts
D01: How much does [service] cost in [location]? D02: How do I choose a reliable [service provider] in [location]? D03: Is it worth hiring a professional for [service or problem]? These test people getting closer to a decision.
Local Prompts
L01: Find a [service provider] in [borough]. L02: Which [service providers] serve [borough]? L03: Recommend a local company for [service] near [area]. These are particularly useful for service businesses working across London boroughs.
Step 4: Keep the Prompts Commercially Relevant
Do not create a prompt bank simply because a question is related to your industry. A pest-control company might appear for:
“What do rats eat?”
That is visibility. But it is a long way from an enquiry. The same business appearing for:
“Who provides rat control in Camden?”
has a different commercial meaning. Your prompt bank should therefore include different stages of the customer journey.
A useful mix is:
- brand prompts,
- problem prompts,
- service prompts,
- local prompts,
- comparison prompts,
and decision prompts. That stops the AI search visibility audit becoming another vanity report.
Step 5: Run the Same Prompts Across Each Platform
Use the same core prompt wording on Google, ChatGPT and Copilot where the platform allows it. Record the date. Record the platform. Record the exact prompt. Do not quietly rewrite the question because you did not like the first answer. If you change the test, you change the measurement.
For Google
Search the query and record whether an AI Overview appears.
If it does, record:
- whether the brand appears,
- whether your site appears as a supporting source,
- which page appears,
and what other sources appear. Also use the dedicated generative AI Search Console report if it is available for the property. Google’s report can show generative AI impressions and break them down by page, country, device and date.
For ChatGPT
Run the saved prompt. Where fresh web information matters, make sure the response actually uses Search.
Record:
- whether the brand appears,
- whether it is recommended,
- whether the details are correct,
- whether the Wixdek or client website is cited,
and which other sources appear. ChatGPT Search can provide web answers with source links, and OpenAI says ranking depends on several factors with no guaranteed top placement.
For Copilot
Run the same prompt. Record the brand mention and citations shown in the response. Microsoft explains that web-grounded Copilot answers use web search and include citations so people can inspect the underlying sources. Then compare the manual observation with Bing Webmaster Tools AI Performance data where available.
Step 6: Create an AI Mention Log
Do not rely on screenshots stored across somebody’s desktop. Put every test into one sheet. A Wixdek AI mention log should contain:
| Field | What to record |
| Date | Date of test |
| Platform | Google, ChatGPT or Copilot |
| Prompt ID | B01, S01, C02 and so on |
| Exact prompt | Wording used |
| Brand mentioned | Yes or no |
| Brand accurate | Accurate, partly accurate or inaccurate |
| Recommended | Yes, no or neutral mention |
| Website cited | Yes or no |
| Cited URL | Exact page used |
| Third-party source | Other source discussing the brand |
| Competitors mentioned | Other providers shown |
| Screenshot saved | Yes or no |
| Referral traffic | Visits where measurable |
| Leads | Leads tied to source where measurable |
| Notes | Anything unusual |
This becomes the evidence base.
Step 7: Score Brand Accuracy
Presence without accuracy can create a problem. Use a simple scale.
Accurate
The important public facts are correct.
Partly Accurate
The business is recognised but one or more details are incomplete or out of date.
Inaccurate
Material information is wrong.
For example:
- wrong location,
- wrong services,
- wrong company type,
- incorrect contact information,
or confusion with another brand. Inaccuracy should receive its own action. Do not celebrate the mention just because the company name appeared.
Step 8: Record Recommendations Separately From Mentions
Suppose ChatGPT says:
“Wixdek is a London SEO agency.”
That is a mention. Now suppose the answer says:
“For a London service business, Wixdek is one provider you could compare.”
That is closer to a recommendation. Keep them separate.
Your audit should use:
- No appearance
- Mention
- Recommendation
Do not call every reference a recommendation.
Step 9: Track AI Citations
Citations help answer a different question:
Which sources are AI systems using when they discuss this topic or business?
Record your own pages. But also record third-party sources.
You may find that the platform knows about the company through:
- an industry directory,
- a news article,
- a review website,
- a partner page,
or another external source. That can reveal how the wider web represents the brand.
Google Citation Measurement
Use real AI Overview observations and Google’s generative AI Search Console reporting. Google’s report measures impressions for URLs appearing in generative AI Search features, though availability was still limited to a subset of sites when Google announced it in June 2026.
ChatGPT Citation Measurement
Record the sources visible in each ChatGPT Search response. Do not assume that being mentioned means your website was the source. OpenAI says ChatGPT Search provides web links and can use third-party search providers when finding current information.
Copilot Citation Measurement
Microsoft gives publishers stronger direct citation reporting through Bing Webmaster Tools. Its AI Performance report shows how often website content is cited, which URLs are referenced and how citation activity changes over time.
Step 10: Track Referral Traffic
AI search visibility becomes easier to value when a user reaches the website. Set up analytics reporting for known AI referral sources. For ChatGPT, OpenAI makes this particularly useful because referral links include the utm_source=chatgpt.com parameter. Your report might include:
| Source | Visits | Engaged visits | Enquiries | Conversion rate |
| ChatGPT | ||||
| Microsoft/Bing AI referrals | ||||
| Google organic and AI Search |
Do not fill gaps with estimates. If a platform does not give you clean referral attribution, say so.
Step 11: Connect AI Referral Visits With Leads
Traffic is still not the end goal.
For a service business, measure whether visitors:
- called,
- completed a form,
- requested a quote,
- made a booking,
or entered the CRM as a genuine lead. If you cannot connect the source to the lead reliably, do not claim that AI search generated it. Label attribution limitations clearly. This is especially important because customer journeys can include several visits before someone enquires. A person may discover your brand in ChatGPT today and search the company name directly next week.
Analytics may not always tell the whole story.
Step 12: Calculate an AI Search Visibility Score
Wixdek can use a simple 100-point model to summarise the audit. This is a Wixdek measurement framework, not an official score provided by Google, OpenAI or Microsoft.
| Area | Weight |
| Brand presence | 20 |
| Brand accuracy | 20 |
| Recommendations | 20 |
| Citations | 15 |
| Referral traffic | 10 |
| Leads | 15 |
| Total | 100 |
The score is useful for comparing the same business over time. Do not use it to claim:
“We have an official AI authority score of 82.”
There is no shared platform metric like that.
Step 13: Repeat the Audit
One audit gives you a snapshot. Several audits give you a trend. Repeat the same prompt set monthly at first.
Keep:
- the prompt wording,
- prompt IDs,
- market,
- platform,
- and scoring system
as consistent as practical. Then compare the results.
Why Repeating Prompts Matters
AI answers are not fixed search-result positions. Google says its AI Overviews and AI Mode may use query fan-out, where related searches are generated to gather information, and the links shown can vary between AI experiences. ChatGPT Search can also rewrite a user’s request into one or more targeted web queries when searching. Copilot can search the web and ground responses using current search results.
Results can therefore change. A business appearing once is interesting. A business appearing consistently across repeated commercial prompts is much more useful evidence.
Do Not Treat One Prompt as a Ranking
Suppose you ask ChatGPT:
“Best SEO agencies London.”
Wixdek appears. It is tempting to write:
“Wixdek ranks number three in ChatGPT.”
Do not do that. An AI answer is not necessarily a stable numbered search result. The wording, sources, context and answer can change.
Record:
- Brand mentioned
- or:
- Brand recommended
rather than inventing a traditional ranking position where one does not exist.
Record Missing Mentions Too
No result is still a result. If your brand does not appear for 15 relevant non-branded prompts, record it. That gives you a baseline. Then repeat the same prompts after improvements. If the business later appears for six, you have evidence of a change in observed visibility. That is far stronger than taking one good screenshot and calling the campaign a success.
How Often Should You Run an AI Search Visibility Audit?
Monthly is a practical starting point for a business actively investing in AI search visibility. Very frequent manual testing can create noise. For a stable business with fewer changes, quarterly reviews may be enough after a baseline has been established. The important point is consistency. Do not test every day until you get the answer you wanted. That is not measurement.
Example: AI Search Visibility Audit for a London Plumber
Imagine a plumbing company operating across Camden and Westminster.
Its fixed prompt set might include:
- What causes low boiler pressure?
- When do I need a plumber for low boiler pressure?
- Recommend an emergency plumber in Camden.
- Which plumbers serve Westminster?
- How much does emergency plumbing cost in London?
- How do I choose a reliable London plumber?
Now test those prompts across the three platforms. The audit may reveal: Google shows the company’s boiler guide as a supporting source for one informational query. ChatGPT does not recommend the company for non-branded plumbing prompts but does describe the brand correctly when asked directly. Copilot cites one of its guides but recommends competing plumbers in the commercial query. Analytics shows two ChatGPT referral sessions but no recorded enquiries.
That is useful information. The result is not simply:
“AI search visibility good.”
The business has informational visibility but weak commercial discovery. That gives the next stage a clear focus.
Example: Wixdek AI Search Visibility Audit
For Wixdek, a fixed prompt bank might test searches around:
- SEO agencies in London,
- SEO consultants,
- technical SEO,
- local SEO,
- SEO pricing,
- AI search optimisation,
and selecting an SEO provider. A baseline could show that Wixdek appears for branded prompts but is less visible when users ask non-branded commercial questions. That does not automatically mean Wixdek needs 100 new blog posts. The next step is to investigate why. Check which sources are being cited. Check whether existing commercial pages match the prompt intent. Check whether important facts are clear.
Review the site’s existing authority and content before creating anything new.
How to Review Brand Accuracy
If an AI system describes your company incorrectly, find the likely source before changing random pages.
Compare the incorrect information with:
- your homepage,
- service pages,
- About page,
- Google Business Profile where relevant,
- trusted third-party profiles,
and pages cited in the AI answer. The aim is to find conflicting or outdated information. Do not manufacture mentions on low-quality websites simply because you want more AI references. Google’s current generative-AI guidance specifically says there is no need to chase inauthentic mentions for its AI Search features.
Should You Add Schema for AI Search Visibility?
Use structured data where it accurately describes visible content and is supported by the search platform. Do not present schema as an AI citation switch. Google explicitly says structured data is not required for its generative AI Search features and there is no special schema markup needed for them. Structured data can still support your wider SEO and search appearance.
Those are different claims.
Do You Need an llms.txt File?
Not for Google Search. Google’s July 2026 generative AI guidance says it does not use llms.txt or similar special AI files for Google Search visibility. Creating one neither improves nor harms Google Search rankings or generative AI search visibility. Other platforms may have different controls. That is another reason not to turn one platform’s guidance into a rule for every AI system.
Do You Need to Rewrite Content Into Tiny AI Chunks?
Google says no. Its current guidance specifically rejects the idea that content must be broken into tiny chunks to be understood by generative Search systems. Short answers can still be helpful. Clear headings can still be helpful. But use them because they improve the page for readers. Not because somebody promised a fixed AI extraction formula.
What Should You Do After the Audit?
Do not start with:
“We need more AI content.”
Start with the evidence.
If the Brand Does Not Appear
Check technical access. Check whether the business is clear on its own website. Review whether the tested prompts genuinely match the services. Look at which competing sources are being used.
If the Brand Appears but Information Is Wrong
Find conflicting or outdated information. Correct important first-party pages. Update genuine external profiles you control. Make the business details clearer.
If the Brand Appears but Is Not Recommended
Review the type of prompt. A purely informational question may not need a provider recommendation. For genuine commercial prompts, compare what recommended businesses provide that your public web presence does not clearly demonstrate.
If Competitors Are Cited More Often
Review the cited pages. Do they offer useful original information? Real examples? Clear service information? Data? Strong explanations? Do not copy them. Use the gap to understand what information your own audience may still need.
If Citations Do Not Produce Visits
Look at intent. A broad informational citation may create awareness without many clicks. Do not rewrite the whole strategy solely to chase click-through rate. Check the wider customer journey.
If AI Visits Do Not Produce Leads
Review the landing pages.
Ask whether the visitor reached:
- an informational guide,
- a commercial service page,
- a location page,
or an irrelevant page. AI search visibility can expose weak conversion paths that were already present.
AI Search Visibility Audit Decision Flow
For the original Wixdek visual, use this sequence in normal text:
- Test fixed prompts
- Then:
- Record brand presence
- Then:
- Check accuracy and recommendation
- Then:
- Record citations
- Then:
- Measure referral visits
- Then:
- Connect visits with leads
- Then:
- Identify the main gap
- Then:
- Make an evidence-led improvement
- Then:
- Repeat the same test
The graphic should show a loop because the audit is repeated. Do not present it as a secret AI ranking algorithm.
Suggested Alt Text
AI search visibility audit framework showing prompt testing, brand mentions, accuracy, citations, referral traffic, leads and repeated measurement.
Wixdek AI Search Visibility Measurement Sheet
The downloadable audit sheet should have four main tabs.
Prompt Bank
Include:
- Prompt ID
- Prompt category
- Exact prompt
- Target service
- Target location
- Customer stage
Mention Log
Include:
- Date
- Platform
- Prompt ID
- Brand appearance
- Accuracy
- Recommendation
- Citation
- Cited page
- Competing brands
- Screenshot
- Notes
Traffic and Leads
Include:
- Platform
- Landing page
- Referral visits
- Engaged visits
- Calls
- Forms
- Bookings
- Qualified leads
Monthly Summary
Include:
- Brand presence rate
- Accuracy rate
- Recommendation rate
- Citation rate
- Referral traffic
- Leads
- Wixdek AI search Visibility Score
This turns the audit into something the team can repeat rather than a one-off research exercise.
Important Limits of AI Search Visibility Measurement
No audit method can remove every variable. AI systems change. Search results change. Models change. Location can affect results. Fresh web information can affect answers. Different accounts and conversations may create different context. This is exactly why repeated testing matters. Your aim is not to pretend AI search visibility is perfectly fixed. The aim is to collect enough consistent evidence to spot useful patterns.
Don’t Compare AI Platforms as Though They Are Identical
Google gives website owners Search Console data. Microsoft now provides AI citation reporting through Bing Webmaster Tools. OpenAI provides web citations to users and referral tracking for publishers, but its publisher guidance does not describe an equivalent Search Console-style impression dashboard. So your report should not pretend every column is equally measurable across every platform. Use:
Measured platform data
where available.
Use:
Controlled manual observation
where it is not.
Label the difference clearly.
AI Search Visibility Is Not a Replacement for SEO Measurement
Keep tracking normal organic search too. Google itself says SEO remains relevant to generative AI Search because those experiences rely on its core Search ranking and quality systems.
Your wider reporting should still review:
- organic visibility,
- commercial queries,
- service-page performance,
- local search,
- calls,
- forms,
- bookings,
and revenue where reliable attribution exists. AI search visibility adds another layer. It does not make the rest disappear.
How This Connects With Google AI Overviews
Our guide on what Google AI Overviews mean for UK service businesses explains Google’s AI Search experience in more detail. The important point for this audit is that Google’s generative AI search visibility can now be measured partly through Search Console rather than relying only on manual screenshots. That makes Google different from a purely manual prompt-tracking workflow.
How SEO, AEO and GEO Fit Into the Audit
You may hear this work described as:
- SEO,
- AEO,
- GEO,
- AI SEO,
or LLM visibility optimisation. Our guide on SEO, AEO and GEO explains the terminology separately. For the audit, the label matters less than the measurement.
Ask:
- Is the business visible?
- Is it represented accurately?
- Is it recommended?
- Is it cited?
- Does anybody visit?
- Do those visits produce business?
Those are much harder questions to hide behind a new acronym.
Make Service Content Easier to Understand
If the audit finds that important pages are unclear or incomplete, the next step may involve improving the content itself. Our guide on how to make service content easier for search engines and AI systems to understand explains that process. The aim should not be to rewrite everything for robots. It should be to make services, evidence, expertise and customer information clearer for everyone.
Common AI Search Visibility Audit Mistakes
The first mistake is testing the brand once. The second is changing the prompt every time. Another is calling a mention a ranking. Some reports count citations but ignore whether the information is correct. Others count visits but never check leads. Another common mistake is assuming every platform uses the same crawler, index or measurement system. And one of the biggest problems is making changes after one strange answer.
AI search visibility should be measured as a pattern. Not a screenshot contest.
How Wixdek Approaches an AI Search Visibility Audit
At Wixdek, the audit should start with a controlled baseline. We define the services and customer journeys that matter. Then we create a fixed prompt bank. The same prompts are tested across Google, ChatGPT and Copilot. We record brand mentions, accuracy, recommendations and citations. Platform data is added where it is genuinely available. Analytics is used to review referral traffic.
Lead tracking is added where attribution can be supported. Then we identify the real gap. Sometimes that gap will be technical access. Sometimes it will be weak commercial pages. Sometimes the business is visible but described badly. Sometimes informational visibility is strong while commercial discovery is weak. And sometimes AI search visibility looks impressive but sends no useful business. That is why the audit needs the full path from mention to lead.
Build a Baseline Before Trying to Improve AI Search Visibility
You cannot measure growth properly if you do not know where you started. Run the prompt set. Save the responses. Build the mention log. Record current citations. Set up referral reporting. Record current leads. Then make changes. When the audit is repeated, compare the new results with the baseline. That is far more credible than saying:
“AI search visibility improved”
because somebody happened to notice your brand in ChatGPT one afternoon.
Get an AI Search Visibility Review From Wixdek
AI search is creating new ways for customers to discover service businesses, but it also creates plenty of noisy metrics. Wixdek can review how your business appears across AI-powered Search alongside your existing SEO performance. Our SEO services in London cover technical SEO, search intent, content, internal linking, authority and conversion-focused search strategy. The aim of an AI search visibility review is not to promise mentions in every AI answer.
It is to establish what can be measured, find genuine visibility gaps and connect search exposure with useful business outcomes.
Request an SEO and AI Search Visibility Review
Frequently Asked Questions
What Is an AI Search Visibility Audit?
An AI search visibility audit checks how a business appears across AI-powered search and answer platforms. It records brand mentions, accuracy, recommendations, citations, referral traffic and leads.
How Do I Check My Brand Visibility in ChatGPT?
Create a fixed set of branded and non-branded prompts, run them through ChatGPT Search and record whether the brand appears, how accurately it is described and whether the website is cited.
Can ChatGPT Referral Traffic Be Tracked?
Yes. OpenAI says ChatGPT automatically adds utm_source=chatgpt.com to referral URLs, allowing publishers to identify ChatGPT search referral traffic in analytics.
What Is OAI-SearchBot?
OAI-SearchBot is OpenAI’s crawler used in relation to ChatGPT search discovery. OpenAI says publishers should allow it if they want their content to be discoverable and cited in ChatGPT search.
Is GPTBot the Same as OAI-SearchBot?
No. OpenAI documents GPTBot separately for potential training controls, while OAI-SearchBot relates to search discovery.
Can Google AI Search Visibility Be Measured in Search Console?
Google launched dedicated generative AI performance reports in Search Console in June 2026. They can show impressions and page-level visibility for supported generative AI Search features, though Google initially rolled them out to a subset of sites.
Can Microsoft Copilot Citations Be Tracked?
Microsoft introduced AI Performance in Bing Webmaster Tools in February 2026. It reports citations across supported Microsoft AI experiences, including Microsoft Copilot, and shows referenced URLs and citation trends.
How Many AI Prompts Should I Track?
There is no official fixed number. The important part is having enough prompts to cover meaningful customer journeys and repeating the same set consistently. This Wixdek framework starts with 18 prompts across six categories.
How Often Should AI Search Visibility Be Checked?
Monthly testing is a practical starting point for an active campaign. The key is to repeat the same prompt set rather than constantly changing the test.
Why Can AI Answers Change Between Tests?
AI search systems can use live or changing search information, rewritten queries and different retrieval methods. Google also says its AI features can use query fan-out to search related topics.
Does Being Mentioned in AI Search Mean My Website Was Cited?
No. A platform may mention a brand without using the brand’s own website as the visible source. Record mentions and citations separately.
Is AI Search Visibility the Same as Leads?
No. Visibility shows that the business appeared. You still need analytics and lead tracking to determine whether that exposure created useful business.
Do I Need Special AI Schema to Improve Visibility?
Google says no special schema is needed for its generative AI Search features. Continue using supported structured data where it accurately represents the page.
Do I Need llms.txt for Google AI Search?
No. Google says it does not use llms.txt for Google Search and that the file does not improve generative AI Search visibility.
