How to Measure AI Search Traffic and Visibility in 2026

AI search reporting has gotten much better in a very short period of time.

Google Search Console now has dedicated reporting for generative AI visibility. Bing Webmaster Tools has an AI Performance dashboard that shows citations and grounding queries. Google Analytics can now separate traffic from recognized AI assistants into its own channel.

That gives website owners several new ways to measure what is happening.

It also creates a new problem.

These platforms are measuring different things.

A citation in Bing is different from an impression in Search Console. An AI Assistant session in GA4 is different from either of those. A lead in your CRM is another step further down the process.

If those numbers do not match, that does not automatically mean anything is broken.

The useful question is what each platform is actually measuring and how the pieces fit together.

There are several different types of AI search data

When someone asks how much AI traffic a website is getting, they may actually be asking several different questions:

• Is my website appearing in AI-generated search results?
• Is my content being cited as a source?
• Which pages are being referenced?
• What topics are those citations associated with?
• Are people clicking through to my website?
• Which AI platforms are sending visitors?
• Are those visitors becoming leads or customers?

There is no single dashboard that answers all of those questions.

I would separate AI reporting into four levels:

1. Visibility
2. Citations
3. Traffic
4. Business outcomes

Each level needs a different data source.

The basic AI reporting stack

A practical reporting setup currently looks something like this:

You may also use third-party AI monitoring platforms or manual searches, but those should usually supplement the first-party data rather than replace it.

Google Search Console can now show generative AI visibility

In June 2026, Google introduced dedicated Generative AI Performance reports in Search Console.

The reports cover visibility in:

• AI Overviews
• AI Mode
• Generative AI features in Discover

Google currently provides data including:

• Impressions
• Pages
• Countries
• Devices
• Dates

The report is being rolled out gradually, so it may not be available in every Search Console property yet.

This is useful because Search Console can finally answer a question that used to require a lot of guessing:

Which pages from my website are appearing in Google’s generative search experiences?

What a Google AI impression actually tells you

An impression tells you that a URL from your site appeared within one of Google’s supported generative AI features.

That is a visibility metric.

It does not tell you:

• Whether the user clicked
• Whether the citation was opened
• Whether your website was the primary source
• Whether the user noticed your source
• Whether the impression produced a lead
• Whether the user later returned through another channel

That distinction becomes important when reporting AI performance.

A page can receive thousands of generative AI impressions while producing very little directly measurable traffic.

Google includes AI impressions in your broader Search Console data

Another important detail is that Google says generative AI performance is still included within the broader Search Console performance totals.

The new report provides a dedicated view of that visibility.

So if your overall Search Console report shows:

100,000 impressions

and the Generative AI report shows:

8,000 impressions

you should not add those together and report:

108,000 impressions

The AI visibility is already represented in the larger total.

The dedicated report is useful for understanding what portion of your visibility came through generative experiences.

Bing Webmaster Tools measures something different

Bing Webmaster Tools now has an AI Performance dashboard.

This is one of the more interesting tools available because it goes further than simply reporting whether a URL appeared.

Bing can currently show:

• Total citations
• Cited pages
• Average cited pages
• Grounding queries
• Citation trends
• Page-level citation activity

Microsoft has also added preview features including:

• Intents
• Topics
• Citation Share
• Period comparisons

The report covers supported experiences such as Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations.

What Bing means by a citation

A citation means your content was visibly referenced or shown as a source in an AI-generated answer.

That is different from an impression.

Bing specifically warns that its citation data does not represent:

• Rankings
• Authority
• Importance
• Traffic
• Clicks
• User engagement

It is measuring citation activity.

That makes Bing’s dashboard useful for understanding how AI systems are using your content.

It should not be treated like a traditional ranking report.

Grounding queries are especially useful

One of the better features in Bing’s AI reporting is its grounding query data.

A grounding query is a phrase associated with the retrieval activity that resulted in your content being cited.

For example, a website about pool construction might see phrases related to:

• pool resurfacing
• pool equipment
• pool remodeling
• swimming pool maintenance

These are not necessarily the exact questions users typed.

Bing describes them as grouped phrases representing retrieval activity rather than complete user prompts.

That makes them useful for discovering what topics your content is being associated with.

Bing can map citations back to specific pages

The Pages view lets you see which URLs receive the most citation activity.

You can also connect grounding queries with cited pages.

That allows you to answer questions such as:

• Which articles are being cited most often?
• What topics are associated with a service page?
• Which pages are appearing across several related grounding queries?
• Which content clusters have the strongest AI visibility?

This can be especially useful for websites with large content libraries.

Bing’s Citation Share adds another useful layer

Citation Share shows your site’s percentage of citations within the citation space for a particular grounding query.

For example:

Grounding query: pool resurfacing
Citation Share: 18%

That does not mean you rank number one for the query.

It means your domain accounted for a portion of the citations Bing observed for that grounding query.

Microsoft explicitly warns that Citation Share is not a ranking or quality score.

Still, it provides a useful way to see whether your presence is increasing or decreasing within an area that matters to the business.

Google Analytics measures actual AI referral traffic

Google Analytics answers a different question.

In May 2026, Google added an AI Assistants channel to GA4.

Google says the channel can identify traffic from sources such as:

• ChatGPT
• Gemini
• DeepSeek
• Copilot
• Grok

Recognized visits receive an ai-assistant medium and appear in the AI Assistants channel within the default channel grouping.

This is much closer to what most people mean when they say “AI traffic.”

These are actual visitors who reached the website from a recognized AI assistant.

Google AI Overviews and AI Mode do not appear in the AI Assistants channel

This is an important distinction.

Google Analytics specifically says that its AI Assistants channel excludes Google’s own AI Overviews and AI Mode.

Clicks from those Google search experiences remain part of Organic Search.

That means a report showing:

AI Assistants: 50 sessions

does not mean your website only received 50 visits influenced by AI search.

Some Google AI-driven visits may be included within Organic Search.

This is one reason AI reporting needs multiple platforms.

GA4 can show which AI platforms actually sent visitors

Once AI Assistant traffic appears in GA4, you can analyze it like other traffic sources.

Look at:

• Sessions
• Users
• Source
• Landing page
• Engagement
• Key events
• Conversions
• Ecommerce activity

You may discover that:

chatgpt.com

sends the most visitors while:

perplexity.ai

sends fewer visitors but produces stronger engagement.

Or you may find that AI referral traffic is still tiny compared with Organic Search.

Both are useful findings.

Referral data still has limitations

GA4 can only classify traffic when it receives enough information to understand where the visitor came from.

If a platform or browser does not pass usable referral information, Analytics may not be able to identify the AI source correctly.

That means AI Assistant traffic in GA4 should be treated as measured AI referral traffic.

It should not automatically be treated as every visit that AI influenced.

A user may:

1. Ask ChatGPT about a company
2. Read the answer
3. Remember the company name
4. Search for the company later
5. Visit through Google

GA4 would likely attribute that visit according to the measurable source that actually delivered the session.

The original AI interaction may never appear in your website analytics.

You can still build custom AI channel groups

Google now provides a default AI Assistants channel, but custom channel groups can still be useful.

GA4 allows custom channel rules, including regular expressions for AI referral sources.

Google even provides an AI assistant example using sources such as ChatGPT, Gemini, Copilot, Claude, and Perplexity.

Custom reporting may be helpful when:

• You want a broader list of AI platforms
• You want to separate individual assistants
• You have older reporting conventions
• You want to compare AI traffic against custom marketing channels
• You need a specific reporting structure for clients

For many sites, the new default AI Assistants channel will be a simpler starting point.

Bing Search Performance also includes Chat activity

Bing’s traditional Search Performance reporting also deserves attention.

Microsoft says its Search Performance report can include clicks and impressions from both Web and Chat sources.

That includes links within Bing Chat responses.

So within Bing Webmaster Tools you effectively have two related views:

Search Performance

This helps show impressions and clicks from Bing search experiences.

AI Performance

This focuses on citations and grounding activity within AI-generated answers.

Those metrics answer different questions.

Why the numbers across these tools will never match

Imagine your reporting looks like this:

Google Generative AI impressions: 12,000
Bing AI citations: 1,300
GA4 AI Assistant sessions: 84
CRM leads attributed to AI: 4

At first glance, those numbers look completely disconnected.

They should.

They are measuring different stages of the process.

A simplified path looks like this:

AI visibility

↓

Source or citation appears

↓

User clicks

↓

Website session is measured

↓

User converts

↓

Lead becomes qualified

↓

Sale occurs

Every step removes some people.

Every platform also has its own reporting rules and coverage.

An AI impression is not a citation

Google’s Generative AI report uses impressions.

Bing’s AI Performance dashboard focuses on citations.

Those should not be compared as equivalent metrics.

An impression generally answers:

Was my URL surfaced?

A citation answers:

Was my content shown as a referenced source?

They are related forms of visibility, but the measurement systems are different.

A citation is not a click

This is another easy mistake.

Suppose Bing reports:

2,000 citations

That does not mean 2,000 people visited the website.

Bing explicitly says citations do not represent traffic or engagement.

The user may have gotten everything they needed from the AI-generated answer.

They may also have clicked another source.

Or they may have clicked your website later through a completely different channel.

A click is not always a measured session

Even when someone clicks, GA4 may not record the visit exactly as expected.

Possible reasons include:

• Consent choices
• Tracking protection
• Ad blockers
• JavaScript failures
• Missing GA4 tags
• Cross-domain behavior
• Redirects
• Referrer loss

That is why Search Console clicks and GA4 sessions have never matched perfectly, even before AI search became a major factor.

A session is not a lead

A visitor arriving from ChatGPT or Copilot may:

• Read one article
• Visit several pages
• Submit a form
• Call the business
• Return later
• Never convert

GA4 can help measure those behaviors where tracking allows it.

The CRM or ecommerce system is still the better place to determine whether the activity produced actual business results.

Measure business outcomes separately

Eventually, AI search reporting needs to connect back to the same things you would measure for any other marketing channel.

Depending on the business, that might include:

• Form submissions
• Phone calls
• Appointments
• Ecommerce purchases
• Quote requests
• CRM contacts
• Opportunities
• Qualified leads
• Revenue

This gives you a much more useful question than:

How many AI citations did we get?

You can ask:

Is increased AI visibility contributing to meaningful business activity?

That is a harder question to answer, but it is the one that actually matters.

Start with Google Search Console

For Google visibility, I would begin with the Generative AI Performance report.

Review:

• Total impressions
• Pages
• Date trends
• Countries
• Devices

Then sort the pages.

Look for:

• Your most visible pages
• Pages gaining visibility
• Pages losing visibility
• Newly visible pages
• Entire content categories performing unusually well

The biggest value is usually in the page-level patterns.

Look for content clusters in GSC

Suppose the top AI-visible pages all relate to:

• Website migrations
• Analytics problems
• CRM integrations
• Technical troubleshooting

That gives you useful information about what Google currently associates with the website.

It may also help guide future content.

Do not immediately create ten more articles about the same topic.

First determine whether the pattern is consistent and whether those topics are commercially useful.

Then review Bing’s AI Performance report

Bing gives you another perspective.

Start with:

• Total citations
• Top cited pages
• Grounding queries
• Citation trends

Then look at the newer features where available:

• Topics
• Intents
• Citation Share
• Compare

The grounding query to page relationship is particularly useful.

It can show you which topics are associated with which pages.

Use Bing intents to understand the type of visibility

Bing currently classifies grounding queries into intent categories that can include:

• Informational
• Commercial
• Navigational
• Comparison
• Research
• Local
• Planning
• Learn and Solve

Microsoft says these labels are generated by AI or machine-learning classifiers and may not always perfectly describe ambiguous queries.

Still, the intent data can be useful.

For a business website, I would care much more about growing citation activity around:

• Commercial
• Local
• Comparison
• Research

than simply accumulating thousands of unrelated informational citations.

Then check GA4

Now move from visibility to traffic.

In GA4, look at the AI Assistants channel.

Review:

• Users
• Sessions
• Engagement
• Landing pages
• Key events
• Conversion rate

Then break it down by source.

You may see traffic from:

• ChatGPT
• Gemini
• Copilot
• Claude
• Perplexity
• Other recognized assistants

The traffic volume may be small.

That does not automatically mean the visibility is unimportant.

AI search is still evolving, and some of the influence happens without a directly measurable referral.

Review AI landing pages

The landing-page report can be particularly useful.

Ask:

• Which pages are AI visitors landing on?
• Are they mostly blog posts?
• Are service pages receiving traffic?
• Are those visitors going deeper into the website?
• Do they convert?
• Are some articles repeatedly driving AI referrals?

You may discover that a page with modest Google search traffic is producing a meaningful amount of AI referral traffic.

That is worth paying attention to.

You may also use third-party AI monitoring platforms or manual searches, but those should usually supplement the first-party data rather than replace it.

Google Search Console can now show generative AI visibility

In June 2026, Google introduced dedicated Generative AI Performance reports in Search Console.

The reports cover visibility in:

• AI Overviews
• AI Mode
• Generative AI features in Discover

Google currently provides data including:

• Impressions
• Pages
• Countries
• Devices
• Dates

The report is being rolled out gradually, so it may not be available in every Search Console property yet.

This is useful because Search Console can finally answer a question that used to require a lot of guessing:

Which pages from my website are appearing in Google’s generative search experiences?

What a Google AI impression actually tells you

An impression tells you that a URL from your site appeared within one of Google’s supported generative AI features.

That is a visibility metric.

It does not tell you:

• Whether the user clicked
• Whether the citation was opened
• Whether your website was the primary source
• Whether the user noticed your source
• Whether the impression produced a lead
• Whether the user later returned through another channel

That distinction becomes important when reporting AI performance.

A page can receive thousands of generative AI impressions while producing very little directly measurable traffic.

Google includes AI impressions in your broader Search Console data

Another important detail is that Google says generative AI performance is still included within the broader Search Console performance totals.

The new report provides a dedicated view of that visibility.

So if your overall Search Console report shows:

100,000 impressions

and the Generative AI report shows:

8,000 impressions

you should not add those together and report:

108,000 impressions

The AI visibility is already represented in the larger total.

The dedicated report is useful for understanding what portion of your visibility came through generative experiences.

Bing Webmaster Tools measures something different

Bing Webmaster Tools now has an AI Performance dashboard.

This is one of the more interesting tools available because it goes further than simply reporting whether a URL appeared.

Bing can currently show:

• Total citations
• Cited pages
• Average cited pages
• Grounding queries
• Citation trends
• Page-level citation activity

Microsoft has also added preview features including:

• Intents
• Topics
• Citation Share
• Period comparisons

The report covers supported experiences such as Microsoft Copilot, AI-generated summaries in Bing, and select partner integrations.

What Bing means by a citation

A citation means your content was visibly referenced or shown as a source in an AI-generated answer.

That is different from an impression.

Bing specifically warns that its citation data does not represent:

• Rankings
• Authority
• Importance
• Traffic
• Clicks
• User engagement

It is measuring citation activity.

That makes Bing’s dashboard useful for understanding how AI systems are using your content.

It should not be treated like a traditional ranking report.

Grounding queries are especially useful

One of the better features in Bing’s AI reporting is its grounding query data.

A grounding query is a phrase associated with the retrieval activity that resulted in your content being cited.

For example, a website about pool construction might see phrases related to:

• pool resurfacing
• pool equipment
• pool remodeling
• swimming pool maintenance

These are not necessarily the exact questions users typed.

Bing describes them as grouped phrases representing retrieval activity rather than complete user prompts.

That makes them useful for discovering what topics your content is being associated with.

Bing can map citations back to specific pages

The Pages view lets you see which URLs receive the most citation activity.

You can also connect grounding queries with cited pages.

That allows you to answer questions such as:

• Which articles are being cited most often?
• What topics are associated with a service page?
• Which pages are appearing across several related grounding queries?
• Which content clusters have the strongest AI visibility?

This can be especially useful for websites with large content libraries.

Bing’s Citation Share adds another useful layer

Citation Share shows your site’s percentage of citations within the citation space for a particular grounding query.

For example:

Grounding query: pool resurfacing
Citation Share: 18%

That does not mean you rank number one for the query.

It means your domain accounted for a portion of the citations Bing observed for that grounding query.

Microsoft explicitly warns that Citation Share is not a ranking or quality score.

Still, it provides a useful way to see whether your presence is increasing or decreasing within an area that matters to the business.

Google Analytics measures actual AI referral traffic

Google Analytics answers a different question.

In May 2026, Google added an AI Assistants channel to GA4.

Google says the channel can identify traffic from sources such as:

• ChatGPT
• Gemini
• DeepSeek
• Copilot
• Grok

Recognized visits receive an ai-assistant medium and appear in the AI Assistants channel within the default channel grouping.

This is much closer to what most people mean when they say “AI traffic.”

These are actual visitors who reached the website from a recognized AI assistant.

Google AI Overviews and AI Mode do not appear in the AI Assistants channel

This is an important distinction.

Google Analytics specifically says that its AI Assistants channel excludes Google’s own AI Overviews and AI Mode.

Clicks from those Google search experiences remain part of Organic Search.

That means a report showing:

AI Assistants: 50 sessions

does not mean your website only received 50 visits influenced by AI search.

Some Google AI-driven visits may be included within Organic Search.

This is one reason AI reporting needs multiple platforms.

GA4 can show which AI platforms actually sent visitors

Once AI Assistant traffic appears in GA4, you can analyze it like other traffic sources.

Look at:

• Sessions
• Users
• Source
• Landing page
• Engagement
• Key events
• Conversions
• Ecommerce activity

You may discover that:

chatgpt.com

sends the most visitors while:

perplexity.ai

sends fewer visitors but produces stronger engagement.

Or you may find that AI referral traffic is still tiny compared with Organic Search.

Both are useful findings.

Referral data still has limitations

GA4 can only classify traffic when it receives enough information to understand where the visitor came from.

If a platform or browser does not pass usable referral information, Analytics may not be able to identify the AI source correctly.

That means AI Assistant traffic in GA4 should be treated as measured AI referral traffic.

It should not automatically be treated as every visit that AI influenced.

A user may:

1. Ask ChatGPT about a company
2. Read the answer
3. Remember the company name
4. Search for the company later
5. Visit through Google

GA4 would likely attribute that visit according to the measurable source that actually delivered the session.

The original AI interaction may never appear in your website analytics.

You can still build custom AI channel groups

Google now provides a default AI Assistants channel, but custom channel groups can still be useful.

GA4 allows custom channel rules, including regular expressions for AI referral sources.

Google even provides an AI assistant example using sources such as ChatGPT, Gemini, Copilot, Claude, and Perplexity.

Custom reporting may be helpful when:

• You want a broader list of AI platforms
• You want to separate individual assistants
• You have older reporting conventions
• You want to compare AI traffic against custom marketing channels
• You need a specific reporting structure for clients

For many sites, the new default AI Assistants channel will be a simpler starting point.

Bing Search Performance also includes Chat activity

Bing’s traditional Search Performance reporting also deserves attention.

Microsoft says its Search Performance report can include clicks and impressions from both Web and Chat sources.

That includes links within Bing Chat responses.

So within Bing Webmaster Tools you effectively have two related views:

Search Performance

This helps show impressions and clicks from Bing search experiences.

AI Performance

This focuses on citations and grounding activity within AI-generated answers.

Those metrics answer different questions.

Why the numbers across these tools will never match

Imagine your reporting looks like this:

Google Generative AI impressions: 12,000
Bing AI citations: 1,300
GA4 AI Assistant sessions: 84
CRM leads attributed to AI: 4

At first glance, those numbers look completely disconnected.

They should.

They are measuring different stages of the process.

A simplified path looks like this:

AI visibility

↓

Source or citation appears

↓

User clicks

↓

Website session is measured

↓

User converts

↓

Lead becomes qualified

↓

Sale occurs

Every step removes some people.

Every platform also has its own reporting rules and coverage.

An AI impression is not a citation

Google’s Generative AI report uses impressions.

Bing’s AI Performance dashboard focuses on citations.

Those should not be compared as equivalent metrics.

An impression generally answers:

Was my URL surfaced?

A citation answers:

Was my content shown as a referenced source?

They are related forms of visibility, but the measurement systems are different.

A citation is not a click

This is another easy mistake.

Suppose Bing reports:

2,000 citations

That does not mean 2,000 people visited the website.

Bing explicitly says citations do not represent traffic or engagement.

The user may have gotten everything they needed from the AI-generated answer.

They may also have clicked another source.

Or they may have clicked your website later through a completely different channel.

A click is not always a measured session

Even when someone clicks, GA4 may not record the visit exactly as expected.

Possible reasons include:

• Consent choices
• Tracking protection
• Ad blockers
• JavaScript failures
• Missing GA4 tags
• Cross-domain behavior
• Redirects
• Referrer loss

That is why Search Console clicks and GA4 sessions have never matched perfectly, even before AI search became a major factor.

A session is not a lead

A visitor arriving from ChatGPT or Copilot may:

• Read one article
• Visit several pages
• Submit a form
• Call the business
• Return later
• Never convert

GA4 can help measure those behaviors where tracking allows it.

The CRM or ecommerce system is still the better place to determine whether the activity produced actual business results.

Measure business outcomes separately

Eventually, AI search reporting needs to connect back to the same things you would measure for any other marketing channel.

Depending on the business, that might include:

• Form submissions
• Phone calls
• Appointments
• Ecommerce purchases
• Quote requests
• CRM contacts
• Opportunities
• Qualified leads
• Revenue

This gives you a much more useful question than:

How many AI citations did we get?

You can ask:

Is increased AI visibility contributing to meaningful business activity?

That is a harder question to answer, but it is the one that actually matters.

Start with Google Search Console

For Google visibility, I would begin with the Generative AI Performance report.

Review:

• Total impressions
• Pages
• Date trends
• Countries
• Devices

Then sort the pages.

Look for:

• Your most visible pages
• Pages gaining visibility
• Pages losing visibility
• Newly visible pages
• Entire content categories performing unusually well

The biggest value is usually in the page-level patterns.

Look for content clusters in GSC

Suppose the top AI-visible pages all relate to:

• Website migrations
• Analytics problems
• CRM integrations
• Technical troubleshooting

That gives you useful information about what Google currently associates with the website.

It may also help guide future content.

Do not immediately create ten more articles about the same topic.

First determine whether the pattern is consistent and whether those topics are commercially useful.

Then review Bing’s AI Performance report

Bing gives you another perspective.

Start with:

• Total citations
• Top cited pages
• Grounding queries
• Citation trends

Then look at the newer features where available:

• Topics
• Intents
• Citation Share
• Compare

The grounding query to page relationship is particularly useful.

It can show you which topics are associated with which pages.

Use Bing intents to understand the type of visibility

Bing currently classifies grounding queries into intent categories that can include:

• Informational
• Commercial
• Navigational
• Comparison
• Research
• Local
• Planning
• Learn and Solve

Microsoft says these labels are generated by AI or machine-learning classifiers and may not always perfectly describe ambiguous queries.

Still, the intent data can be useful.

For a business website, I would care much more about growing citation activity around:

• Commercial
• Local
• Comparison
• Research

than simply accumulating thousands of unrelated informational citations.

Then check GA4

Now move from visibility to traffic.

In GA4, look at the AI Assistants channel.

Review:

• Users
• Sessions
• Engagement
• Landing pages
• Key events
• Conversion rate

Then break it down by source.

You may see traffic from:

• ChatGPT
• Gemini
• Copilot
• Claude
• Perplexity
• Other recognized assistants

The traffic volume may be small.

That does not automatically mean the visibility is unimportant.

AI search is still evolving, and some of the influence happens without a directly measurable referral.

Review AI landing pages

The landing-page report can be particularly useful.

Ask:

• Which pages are AI visitors landing on?
• Are they mostly blog posts?
• Are service pages receiving traffic?
• Are those visitors going deeper into the website?
• Do they convert?
• Are some articles repeatedly driving AI referrals?

You may discover that a page with modest Google search traffic is producing a meaningful amount of AI referral traffic.

That is worth paying attention to.

You do not need perfect numbers.

You are looking for patterns.

A page performing across multiple systems is especially interesting.

High visibility with little traffic

This is likely to become common.

You might see:

• Large AI impression volume
• Frequent citations
• Very few referral sessions

That can happen when AI answers satisfy most of the user’s question directly.

It may also happen because the AI interface does not send easily measurable referral data.

Look at the business context before deciding whether this is good or bad.

Low visibility with strong AI referral traffic

This can happen too.

A website may receive meaningful traffic from ChatGPT or another assistant even though Google’s or Bing’s first-party visibility dashboards show little activity.

Those systems only cover their own supported experiences.

ChatGPT traffic, for example, would not be represented by Google’s generative AI impression report.

This is another reason a single dashboard cannot represent overall AI performance.

AI traffic can convert differently from search traffic

Do not assume AI visitors behave exactly like Google search visitors.

Compare:

• Engagement rate
• Pages per visit
• Lead rate
• Purchase rate
• Average order value
• Time to conversion

Users coming from an AI assistant may already have received substantial information before clicking.

They may arrive farther along in the decision process.

Or they may simply be gathering more information.

Your own data will tell you more than broad industry averages.

Check your CRM

This is where the measurement can become commercially useful.

If your forms capture source data or UTMs, review whether AI-originated leads are entering the CRM.

Then look at:

• Lead quality
• Assigned salesperson
• Opportunity creation
• Close rate
• Revenue

You may eventually discover that AI referral traffic is small but produces unusually strong leads.

You may also discover that it produces a lot of informational visits with almost no commercial outcome.

Both findings are useful.

Ask leads how they found you

There is still value in simply asking.

A form field such as:

How did you hear about us?

may reveal answers like:

• ChatGPT
• Google AI
• Copilot
• Perplexity
• AI search
• Google

Self-reported attribution is imperfect.

People forget where they found a company.

They may describe Google AI Mode simply as Google.

Still, it provides another source of evidence when combined with analytics.

Manual AI searches can still be useful

First-party dashboards have improved substantially, but manual testing still has a role.

Search the questions that matter to the business and see:

• Whether AI Overviews appear
• Which sources are cited
• Whether your company is mentioned
• Whether competitors are cited
• Which type of content appears

Do not treat a handful of manual searches as a statistically valid visibility report.

AI results can vary based on:

• Query wording
• Location
• Personalization
• Search history
• Model changes
• Time

Use manual testing as supporting evidence.

Third-party AI visibility tools can add another layer

There are now many tools that attempt to monitor visibility across:

• ChatGPT
• Gemini
• Perplexity
• Copilot
• Google AI Mode
• Other LLMs

These tools can be useful for:

• Prompt monitoring
• Competitor comparisons
• Brand mentions
• Citation tracking
• Share-of-voice estimates

The limitation is that they generally test a selected set of prompts.

That means the report reflects:

How you performed across the prompts the tool tested.

It does not represent every AI query that could potentially surface your content.

That distinction should be clear when reporting results.

Do not report all AI metrics together as one number

I would avoid a dashboard that says:

AI Visibility: 18,472

after adding together:

• Google impressions
• Bing citations
• GA4 sessions
• Third-party mentions

Those numbers measure different things.

Adding them produces a number that has no real meaning.

Keep the categories separate.

A better monthly AI report

I would organize the report into sections.

Google generative search visibility

Report:

• AI impressions
• Change from previous period
• Top visible pages
• New visible pages

Microsoft and Copilot citations

Report:

• Total citations
• Cited pages
• Top grounding queries
• Important intents
• Citation Share where useful

AI referral traffic

Report:

• AI Assistant sessions
• Top AI sources
• Top landing pages
• Engagement
• Conversions

Business results

Report:

• Leads
• Calls
• Purchases
• Opportunities
• Revenue

Then add a short interpretation.

For example:

AI visibility increased during the month, led primarily by three educational pages. Bing citation activity also increased around several related topics. Direct AI referral traffic remains relatively small, but visitors from AI assistants showed strong engagement and generated two qualified leads.

That gives the client useful context.

Track changes over time

AI reporting is much more useful as a trend.

Compare:

• Month over month
• Quarter over quarter
• Before and after important content changes

Document major events such as:

• New articles
• Content updates
• Website redesigns
• Internal linking changes
• Technical SEO work
• Migrations
• Schema changes
• New service pages

Then watch whether visibility changes.

Be careful about claiming causation.

Microsoft specifically notes that citation changes can result from user demand, model updates, content changes, partner refresh cycles, and changes across the web.

The same caution should apply to AI visibility generally.

Look for pages appearing across several systems

One of the strongest signals available right now is consistency.

Suppose one article:

• Earns Google generative AI impressions
• Receives frequent Bing citations
• Appears for useful grounding queries
• Gets ChatGPT referral traffic
• Generates leads

That deserves attention.

The individual metrics may be imperfect.

Together, they provide much stronger evidence that the content is useful across several discovery systems.

Look for commercially useful AI visibility

High AI visibility is not automatically valuable.

A business can become highly visible around topics that never produce meaningful customers.

Pay attention to:

• Commercial topics
• Service-related questions
• Product questions
• Comparison queries
• Local intent
• Research queries that occur before a purchase

This is where Bing’s intent reporting may become especially useful.

A smaller amount of visibility around high-value commercial topics may matter more than a huge number of citations around general information.

Do you need a separate AI SEO strategy?

I would be careful with that framing.

AI search has created new reporting systems, new interfaces, and new opportunities for websites to be cited.

The underlying work still overlaps heavily with good technical SEO and useful content.

That includes:

• Crawlable pages
• Proper indexing
• Clear site architecture
• Strong internal linking
• Accurate information
• Useful explanations
• Original experience
• Well-supported claims
• Clear headings
• Appropriate structured data
• Fresh content

Google’s new Generative AI Performance report is part of Search Console, and Microsoft’s AI Performance guidance similarly recommends useful, clear, well-structured, current content rather than a separate technical trick.

Do you need special schema for AI search?

There is currently no special Google schema markup required for appearing in AI Overviews or AI Mode.

Structured data can still help Google understand content when it accurately represents what is on the page.

Use schema because it properly describes the content and supports relevant search features.

Avoid adding unsupported markup simply because someone claims it is required for AI visibility.

What about llms.txt?

You may also see recommendations to add an llms.txt file specifically for AI systems.

That standard has received attention, but it should not be treated as a replacement for:

• robots.txt
• XML sitemaps
• Search Console
• Bing Webmaster Tools
• Internal linking
• Crawlable HTML
• Proper indexing

For Google and Bing visibility, the first-party tools and normal crawl/index controls are much more important to monitor.

Build content around actual evidence

AI reporting is finally giving us more information about what content these systems use.

That means content strategy can become more evidence-based.

If you find that a group of pages consistently earns:

• Google AI impressions
• Bing citations
• Relevant grounding queries
• AI referral traffic

look at what those pages have in common.

Maybe they:

• Answer narrow technical questions
• Explain troubleshooting procedures
• Include firsthand experience
• Cover subjects deeply
• Use clear headings
• Define confusing concepts
• Give specific examples
• Cite reliable sources

Those patterns can help guide future content.

Avoid blindly copying a format just because one page performed well.

AI visibility can grow without traffic growing

This may become one of the biggest reporting challenges.

Imagine:

Google AI impressions: +70%
Bing citations: +45%
AI referral sessions: +8%
Organic traffic: -5%
Qualified leads: +12%

That is a complicated result.

Calling it either an SEO success or failure based on one metric would miss most of the picture.

Now imagine:

AI impressions: +70%
AI citations: +45%
Website traffic: -30%
Qualified leads: -35%

That deserves considerably more concern.

The value comes from looking at the full path.

A practical AI measurement workflow

For most websites, I would use this process.

1. Check Google Search Console

Review generative AI impressions and the pages receiving visibility.

2. Check Bing Webmaster Tools

Review citations, grounding queries, pages, intents, topics, and Citation Share where available.

3. Check GA4

Review AI Assistants, sources, landing pages, engagement, and conversions.

4. Check actual website conversions

Review:

• Forms
• Calls
• Ecommerce
• Booking systems

5. Check the CRM

Determine whether the traffic produced meaningful leads or opportunities.

6. Compare the pages

Identify URLs performing across multiple systems.

7. Document recent changes

Record new content, updates, internal linking, migrations, or technical work.

8. Watch the trends

AI reporting is still evolving quickly.

A month-over-month pattern is much more useful than one screenshot.

What should you include in an AI SEO report?

For most businesses, I would keep the report relatively simple.

Include:

Google

• Generative AI impressions
• Top pages
• Change over time

Bing / Copilot

• Citations
• Cited pages
• Important grounding queries
• Useful intent or topic trends

GA4

• AI Assistant sessions
• Major AI sources
• Top landing pages
• Conversions

Business outcomes

• Leads
• Purchases
• Calls
• Opportunities

Then explain what changed and what you think deserves attention.

The explanation is usually more valuable than another graph.

Do not promise more precision than the data provides

AI search reporting has improved dramatically.

It is still incomplete.

You cannot currently determine every time an AI system influenced a purchase.

You cannot reliably connect every generative AI impression to a website session.

You cannot assume every citation had equal visibility.

You cannot assume every person who heard about a company through AI clicked directly from that AI platform.

Good reporting should make those limitations clear.

That gives the numbers more credibility.

When AI reporting deserves a deeper technical review

A deeper review may make sense when:

• AI visibility is changing significantly
• Search Console and GA4 seem to tell very different stories
• Bing citations increase but traffic does not
• AI referral traffic appears in unexpected channels
• Important pages receive little visibility
• A redesign or migration coincided with a major change
• Conversion tracking is unreliable
• Several analytics implementations are present
• Nobody knows which platform should be trusted
• The business wants to establish a reliable AI search baseline

Kismet’s Technical Website Assessment can review Search Console, GA4, website architecture, indexing, tracking, and technical SEO signals to help determine what the available data actually supports.

—

Understand what AI search is actually doing for your website

AI search now shows up in several reporting systems.

Each one provides a different part of the picture.

Google Search Console can show visibility. Bing Webmaster Tools can show citation activity. GA4 can show measurable AI referral traffic. Your CRM, forms, ecommerce platform, and sales data can tell you whether any of it produced a business result.

Putting those pieces together gives you a much more useful view of AI search performance than relying on any single dashboard.

Request a Technical Assessment

—

Written by Joey Zuccarini
COO at Kismet Creative Co.