GA4 CSV analyser
Find the AI traffic GA4 buries.
Upload a Traffic acquisition export to separate identifiable ChatGPT, Claude, Gemini, Perplexity, Copilot, and other AI referrals, then compare sessions, engagement, key events, landing pages, and daily movement.
Your data stays on this device
Files are read in your browser and are not sent to Trakkr. Refreshing or resetting clears the analysis. Analytics record only the tool name, input method, file-size band, and row count, never file names or contents.
Add a GA4 export
Required: Session source and Sessions. Recommended: Date, Engaged sessions, Key events, and Landing page + query string.
Method and limits
Known referrers
High-confidence hostnames are grouped by platform. Broad values such as openai and anthropic stay marked for review.
No dark-traffic guess
Direct and unassigned traffic remain unclassified. The tool never invents a hidden AI number from a change in Direct traffic.
Human sessions only
GA4 measures visits where browser analytics ran. AI crawler requests belong in server or CDN logs, not this report.
Keep measuring
Connect traffic to visibility
Trakkr combines AI referral analytics with the prompts, citations, rankings, and crawler activity that explain why traffic changed.
Common questions
Which GA4 report should I export?
Use Traffic acquisition with Session source as the primary dimension. Add Date, Landing page + query string, Sessions, Engaged sessions, and Key events, then export every source rather than filtering to AI first.
Can this find dark AI traffic?
No. If an app, redirect, privacy setting, or referrer policy strips the source, GA4 may classify the visit as Direct. The analyser reports identifiable AI referrals as an observed minimum and does not relabel Direct traffic.
Does Google AI Overview traffic appear here?
Not reliably. Google AI Overview and AI Mode clicks can remain google / organic. Gemini app referrals may appear as gemini.google.com, but that is a different surface.
Is the CSV uploaded?
No. The CSV is read and classified in your browser. Trakkr does not receive the source rows, landing pages, conversions, file name, or property details.