AI Search and Customer Inquiries: What Can You Measure?

AI visibility, website visits, and customer inquiries are three different things.

Mix them into one number and the report can look more impressive while telling you less.

A business can appear in an AI answer without receiving a click. Somebody can visit the website without contacting the business. An inquiry can arrive without any reliable record of what influenced the person earlier.

So before deciding whether AI search deserves more time or budget, separate the questions.

Start with four business questions

A useful AI-search review can begin with four ordinary questions:

1. Are we appearing in relevant AI or search answers?
2. Are people reaching our website from sources we can identify?
3. Are they contacting us?
4. Are those inquiries actually suitable for the business?

Those are four stages, not one guaranteed funnel.

A prospect might discover the business in an assistant, click through, and make an inquiry immediately. Another person might see the business in an AI answer, remember the name, and search for it later. Many people will never visit at all.

Our guide to mentions, citations and referrals covers the more formal observation method. Here the focus is simpler: what can a business owner reasonably learn from the evidence, and what should happen next?

Do not count crawlers as customers

Search and AI companies use automated crawlers and fetchers to read websites.

That activity can tell you that a system accessed the site. It does not tell you that a person considered the business, visited the website, or sent an inquiry.

OpenAI, for example, documents different bots for search, training, and user-triggered fetching. Those purposes matter operationally, but none of them turns a crawler request into a lead. [1]

What the current reports can tell you

The reporting is getting better, but there is still no single “AI number” that answers everything.

Google now documents a Generative AI performance report in Search Console. It can show impressions in AI Overviews and AI Mode when the report is available and there is enough data. That is visibility evidence, not a count of customer inquiries. Google also says this data is already included in the broader Web performance data, so the totals should not simply be added together. [2]

Bing has introduced AI visibility reporting that includes Citation Share, which describes how often a site is cited for a particular grounding query. Microsoft explicitly says this is not traffic share and not a content-quality score. [3]

GA4 can help with visits, but its labels need to be read carefully. Google’s current default channel rules include an AI Assistant channel for recognized assistant sources such as ChatGPT and Gemini. Google’s own AI Overviews and AI Mode are not placed in that channel; they are included under Organic Search. [4]

In other words, “AI Assistant traffic” in GA4 is not the same thing as “all traffic influenced by AI.”

For a business owner, the important point is not memorizing the channel rules. It is knowing what the report does and does not include before making a budget decision.

Connect observable visits with real inquiries

OpenAI documents utm_source=chatgpt.com in referral links from ChatGPT. When that information reaches the site, it can provide a useful observable source. [6]

But even a clean referral cannot show everything that happened before the click. And sometimes the customer may return later through Google, a bookmark, or a direct visit.

That is why the inquiry record matters.

Define the business outcomes before comparing channels:

• accepted inquiry — a real request that reached the business;
• qualified inquiry — a request that fits the services and basic criteria;
• won project / customer — actual business outcome.

Those are not interchangeable.

For a local service, qualification may depend on the type of work and service area. For a professional project, it may depend on scope and whether there is a meaningful next conversation.

Use the business’s real criteria. Do not upgrade a weak inquiry just because the source label says AI.

Keep what analytics saw separate from what the customer says

Suppose a prospect says:

I first found you in ChatGPT.

But the inquiry record shows that the final visit came from Google.

There is no need to “fix” one of those facts.

The first tells you what the customer remembers about discovery. The second tells you what the website recorded about the later arrival. Both can be useful if they are kept separate.

Likewise, a rise in Direct traffic or branded searches can be a reason to investigate. It is not proof that AI caused the increase.

And a month with no clearly identifiable AI inquiries does not prove that AI had no influence. Sometimes the honest answer is simply: we cannot reliably assign the earlier influence from the available data.

That sentence is more useful than a precise-looking number built from guesses.

Let the gap decide the next task

The most useful report ends with a decision.

AI/search describes the business incorrectly
Check the confirmed services, locations, company facts, and supporting pages available to customers and search systems. Google’s current guidance still emphasizes useful original content and clear technical access rather than a special AI markup trick. [7]

Google AI visibility appears weak
Check indexing, whether Google can use the page in search features, and the current inclusion controls before assuming the answer is “publish more content.” [8]

Relevant visits arrive, but suitable inquiries do not
Look at the landing page. Does it make the service fit clear? Is the evidence convincing? Is the next step obvious?

Inquiries arrive, but the source is usually missing
Fix the measurement before claiming an AI return on investment.

The IRFĒ client story includes a saved AI Mode source-panel example alongside historical search evidence. It shows an observed appearance in AI search. It does not claim that AI produced sales, and it should not be used to promise the same result for another business.

Ask for a business report, not a bigger-looking total

A useful report keeps these separate:

• discovery observations;
• identifiable website visits;
• accepted inquiries;
• qualified opportunities or customers;
• anything that remains unknown.

It should state the period, the source of each number, and the limits of the attribution.

Then it should name the next problem to solve.

Our AI search work focuses on discovery and the evidence search systems can find about a business. Our analytics and conversion work focuses on what can be observed after the visit and where the inquiry journey may be breaking down.

The goal is not to make the AI column larger. It is to know enough to make the next business decision without pretending the data says more than it does.

Sources

1. OpenAI crawler documentation:https://developers.openai.com/api/docs/bots
2. Google Search Console — Generative AI performance report:https://support.google.com/webmasters/answer/16984139
3. Bing — AI visibility / Citation Share announcement:https://blogs.bing.com/search/2026/6/New-AI-Visibility-Insights-in-Bing-Webmaster-Tools-Intents-Topics-Citation-Share-Compare/
4. Google Analytics — Default channel group definitions:https://support.google.com/analytics/answer/9756891
5. Google Analytics — Traffic acquisition report:https://support.google.com/analytics/answer/12923437
6. OpenAI publisher/developer FAQ — referral URLs:https://help.openai.com/en/articles/12627856-publishers-and-developers-faq
7. Google generative AI optimization guidance:https://developers.google.com/search/docs/fundamentals/ai-optimization-guide
8. Google Search generative AI controls:https://support.google.com/webmasters/answer/16908024

Back to Blog