Analytics
How to Check AI Referral Traffic in Your Analytics
If an AI-service domain appears as the source, analytics has recorded visits attributed to that domain. UTM tags help identify visits associated with a tagged link; they do not prove that an AI service placed it. Direct traffic does not establish a source.
An AI service mentioning your website does not guarantee that your analytics will record a visit from it. Analytics can identify a source when referral or campaign information reaches the system and is processed. Some visits may be attributed to an AI-service domain; others may be associated with links you tagged. The source of the rest may remain unknown.
In brief: check the recorded source. If an AI-service domain appears, you can report visits attributed to that domain. UTM tags can help identify visits associated with a tagged link, but they do not show where a visitor saw it. Direct traffic does not confirm an AI referral.
What the source data can establish
If an AI-service domain appears as the source, you can report that analytics recorded visits attributed to that domain during the selected period. The report does not show which answer or mention prompted a visit, and it does not count every appearance of your business in AI-generated answers.
Campaign parameters provide a different signal: they help you find visits for which analytics processed values you added to a link. This can be useful for a link in your own newsletter or a controlled test. Tags do not show where a visitor saw the link unless you controlled its placement.
Referral reporting can help you assess visits for which analytics identified a source. It does not measure your overall visibility in AI search or reveal a source that the system did not receive.
Where to check in Google Analytics 4
- Select the property and date range. Make sure the report covers the right website and dates. Record the period for your report.
- Open a Traffic acquisition report or an Exploration. Check session-level dimensions such as Session source / medium. If you use UTM tags, check Session campaign as well.
- Review the source values. If an AI-service domain appears, record it and the number of sessions shown for the selected period.
- Write down your inclusion rule. List the domains you count as AI services. Do not group rows by channel name alone if that does not identify the source.
Google explains how Analytics collects and processes referral and campaign information to populate traffic-source dimensions. The data may include a page referrer and campaign parameters. See Google’s documentation on campaigns and traffic sources.
Google defines (direct) / (none) as traffic without a clear referral source. Possible causes include someone entering a web address directly, following a link in an offline document or losing source information through a redirect. The category does not identify which explanation applies to a particular visit, so it cannot confirm an AI referral. See Google’s guidance on direct traffic.
Where to check in Yandex Metrica
Review the traffic-source summary, referring-sites and UTM-tag reports. If an AI-service domain appears among the referral sources or in the referring-sites report, record it along with the report name and date range. Report names may vary with the interface language.
Yandex says Metrica identifies traffic sources using the HTTP Referer header and tags. It checks tags using data from the first page view in a visit; if that view has no tags, it uses the referrer. If the referrer does not pass the source and there are no suitable tags, a visit may be classified as direct. These are Yandex Metrica rules, not Google Analytics 4 instructions. See Yandex Metrica’s guide to identifying traffic sources (in Russian).
If a source you expected to see is missing, check whether the link passed through a redirect and whether its campaign parameters remained in the URL. If the report contains no evidence that identifies the source, do not infer it from the landing page or a separate mention of your website.
When UTM tags help
Add tags to links whose URLs you control, such as links in your own newsletter or a defined test campaign. Use consistent source, medium and campaign values, then check whether they appear in your reports.
For example, a test campaign might use ?utm_source=owned_campaign_test&utm_medium=referral&utm_campaign=autumn_review. This illustrates a naming scheme; it is not a universal standard for tracking AI referrals. The tags can help you find visits associated with that link, but they do not prove that an AI service published it or that a visitor saw it in an AI-generated answer.
If an AI service or another publisher links to your website and you do not control the URL, you cannot add tags to it after the fact. Use the referral information your analytics actually recorded. If no source information is available, leave the source unknown.
After tagging a link, check its final destination and confirm that the parameters survive any redirects. Then check the relevant fields in your report. Google notes that redirects and link shorteners can result in missing referral or campaign information.
What each signal supports
| What the report shows | What you can conclude | What it does not prove |
|---|---|---|
| An AI-service domain appears as the source | Analytics recorded visits attributed to that domain | Which answer, mention or link prompted a visit |
| Your UTM tags appear for a visit | Analytics processed the tag values you set for the link | That an AI service placed the link or included it in an answer |
The source is direct or (direct) / (none) | The system did not identify a clear source from the available data | That the visit came from AI search or any other specific channel |
| A visitor landed on a page related to AI search | The page received a recorded visit | Which source brought the visitor there |
For example, traffic to a page may rise during the same period that your brand appears in AI answers. If no source was recorded, that timing alone does not connect the visits to those mentions.
How to report the result accurately
State which domains and tags you included, which analytics property you used and the date range. For example: “During the stated period, Analytics recorded sessions with sources from the listed domains.” Add figures only after checking the report, and explain the criteria you used.
Say separately when the source of some visits could not be identified. Do not add those visits to a confirmed total or include direct traffic on the assumption that it came from AI answers. If you combine known and unknown visits for an internal estimate, label the result as an estimate and explain the assumption.
When comparing periods, use the same property, date-selection rules, domain list and inclusion criteria. If any of these change, note it beside the figures so readers can tell whether the traffic changed or the counting method did.
Check before you share the report
- Record the analytics property, report and date range.
- List the domains and UTM values used to select visits.
- Check that your tags appear in the intended fields and survive redirects.
- Keep visits with an identified source separate from direct visits and visits with an unknown source.
- Describe the figures as identified referrals, not as every mention of your business in AI services.
Assess visibility separately from referrals
Web analytics reports visits whose source the system could identify. It does not show every instance in which a business appears in an AI service’s answers. A referral report is therefore not a measure of overall visibility.
An AI search visibility audit addresses the separate question of whether and where a business appears in AI search. For broader guidance on interpreting performance measures and reports, see How to measure SEO results: metrics and reporting.
Share this article