How Local Businesses Can Measure Traffic From AI Search
Separate identifiable referrals, citations and conversions. Build a monthly report that follows the enquiry instead of celebrating a mention.
By Sarah Scarano · Published
A screenshot of an AI assistant naming your business is not an acquisition report. It proves that a particular answer named the business at a particular moment. A visit, an enquiry and a sale need their own evidence.
Start with three columns: observed citations, identifiable referral visits, and completed customer actions. Keeping them separate makes the numbers useful.
Recognize the source without overstating it
Visits can carry a referrer from domains such as chatgpt.com, perplexity.ai, gemini.google.com or copilot.microsoft.com. A tagged link can also contain a source parameter. Use a strict hostname comparison. A domain that merely contains the word “chatgpt” is not proof of a ChatGPT referral.
Record which evidence you have. A recognized referrer and a user-supplied UTM tag are different signals. Anyone can write a tagged URL, including your own team testing the site. Exclude those tests from commercial reporting.
Do not force unknown traffic into an AI bucket. Apps, privacy settings, redirects and copied links can omit referral information. Those visits remain unclassified unless another reliable source identifies them.
Find the traffic in GA4
In a configured GA4 property, open Reports, Acquisition, Traffic acquisition and use Session source / medium. Filter for the recognizable source domains and inspect the landing pages and completed events. Google's traffic-acquisition documentation explains the session-based report and its dimensions.
The report is about visits. It cannot count every person who saw a recommendation and did not click. A zero in a referral report does not establish zero AI visibility.
For MMS, the implementation adds an AI source to existing analytics events after the visitor grants analytics consent. The source persists through the browsing session, so a visitor can arrive on a guide and later book a call without losing that context. With consent declined, the analytics events and that storage remain off.
Follow the action that matters
A useful funnel is landing page, demo or contact entry, successful enquiry or booking, then the commercial result recorded by the business. Track success after the operation succeeds. A click on “Book” is interest; a confirmed appointment is a booking.
MMS already has lead, booking and purchase events. Attaching source context to those existing events is more useful than inventing another dashboard that counts only visits. It also avoids treating a failed form submission as a lead.
Campaign parameters should survive important internal landing-to-conversion links where the implementation depends on them. Preserve the destination's own parameters and fragments. Do not decorate external, payment, authentication or signed URLs indiscriminately.
Build the monthly sheet
| Measure | Evidence | What it establishes |
|---|---|---|
| Business mentioned | Exact prompt, answer, system, date | One observed answer |
| Page cited | Citation URL and supported claim | A source used in that answer |
| Identifiable AI visit | Referrer or documented campaign source | A classified visit, subject to available attribution |
| Enquiry or booking | Success event and saved record | A completed action |
| Customer outcome | CRM or payment record | The business result actually recorded |
Review landing pages as well as totals. If a guide attracts visitors who never reach a relevant offer, inspect its next step. If visitors start a demo and fail at the same field, fix the flow before publishing another guide.
Benchmark questions consistently
Repeat a short set of customer questions each month. Record search-enabled versus non-search answers separately. Keep the phrasing, location and observation method stable. Save cited sources and factual errors, not only whether the business appeared.
Use the log to decide what evidence is missing. Then improve the page that should answer the question. Our Montana guide and AI website explanation show how those corrections connect to a useful customer journey.
The report should end with a decision: fix a broken handoff, improve a specific service page, substantiate a claim, or continue observing. A chart that cannot change the next move is decoration.
Put the field notes to work.
Explore our AI websites, see a free demo, or talk with Sarah about your business.
- The Montana Business Owner's Guide to Getting Found by ChatGPT
- How ChatGPT Finds Local Businesses: What Owners Can Verify
- GEO vs SEO: What Montana Businesses Actually Need