Ask callers how they found you, and watch what your website analytics show
The fastest way to tell if AI search is sending you windshield customers is to ask every caller a simple question: "How did you hear about us?" Pair that with a look at your website's referral traffic for visits coming from AI assistant domains. Between the two, a pattern emerges within a few weeks that tells you whether tools like ChatGPT, Gemini, or Perplexity are already part of your booking pipeline.
Why AI referrals are harder to see in analytics
AI search referrals do not show up the way a Google search click does. When someone asks ChatGPT or Gemini for an auto glass shop and then visits your site, standard web analytics tools often lump that visit under "direct traffic" or fail to log a source at all, because the AI assistant does not always pass referral data the way a search engine results page does. This makes AI-driven traffic look invisible even when it is happening, which is why phone and counter conversations matter as much as any dashboard.
Simple intake questions that surface AI-referred callers
A short, consistent intake question at the point of contact catches what analytics miss: ask "How did you hear about us?" on every call, text, and walk-in, and log the answer in whatever system you already use to schedule jobs. Listen for phrases like "I asked ChatGPT," "an AI search told me," or "I asked my phone assistant," since customers rarely say "AI search" unprompted but will describe the tool by name if asked a follow-up.
Train whoever answers the phone to treat this as a two-part question, not a checkbox. The first answer might be vague, such as "I searched online." A quick follow-up, "Was that a regular Google search or did you ask an AI tool like ChatGPT?," gets a clearer answer without sounding like an interrogation. Over time, this single habit builds a record that no analytics platform can replicate, because it captures the moment a customer decided to call, not just the moment they clicked a link.
Keep the log simple: a spreadsheet column or a note field in your booking software works fine. What matters is consistency across every staff member who takes calls, so the data reflects the whole shop rather than one attentive employee's memory.
What to watch in website referral data
Website referral data fills the gap that phone logs cannot cover, especially for customers who research before calling. Check your analytics for traffic sources tied to AI assistant domains, and pay attention to any spike in visits to specific pages, like your windshield replacement pricing page or your service area list, since AI assistants often send users to a single answer page rather than your homepage.
Look for landing page patterns rather than just source labels. If a page about mobile windshield repair or ADAS (advanced driver assistance systems, the cameras and sensors mounted near many modern windshields) calibration suddenly gets more direct or unlabeled traffic without a matching increase in social or ad activity, that page may be getting surfaced in AI-generated answers. Cross-reference the timing with your intake log. If callers who mention an AI tool tend to ask about the same service, that is a strong sign the two data sources are describing the same trend from different angles.
Do not expect a clean, single-source report. AI referral tracking is still uneven across platforms, and no analytics tool currently offers a complete, guaranteed breakdown of AI-driven visits. Treating the website data as a supporting signal, not a definitive count, keeps expectations realistic while still giving you something to act on.
Using the findings to guide your next content
Once intake questions and referral data start showing a pattern, that pattern should shape what you publish and update next. If callers mention asking an AI tool about windshield replacement cost, insurance claims, or same-day mobile repair, those are the exact questions your website content should answer clearly and directly, since AI assistants tend to pull from pages that already state a plain answer near the top.
If the data shows AI-referred customers concentrated around one service, like rock chip repair or ADAS calibration, expand that page with the specific details drivers ask about: what the process involves, how long it takes, and what determines whether a chip can be repaired versus requiring full replacement. If the pattern shows almost no AI-referred callers yet, that is useful too. It means your near-term focus should stay on making sure your basic service pages, hours, and locations are easy for any search tool to find and quote accurately, so you are positioned once that traffic starts arriving.
Revisit both the intake log and the referral data on a regular schedule, such as monthly, rather than treating this as a one-time check. AI assistants change how they source and phrase answers over time, and a shop that keeps watching will notice shifts in customer language before competitors do.
The clearest sign that AI search is already sending you windshield customers is not a single metric but the overlap between what callers say and what your website data shows. When intake answers naming an AI tool line up with traffic spikes on the same service pages, that overlap is the real signal, and it tells a shop exactly which questions to answer better next.