Definition
AI search traffic is the visitors who reach your site from a link inside an AI assistant’s answer. Unlike organic search traffic, which carries a referrer you can read in analytics, many of these visits arrive with the referrer stripped and land in the direct bucket. The measured number is always lower than the real one.
At a glance
- CategoriesMeasurement
- Related fieldsAnalytics
- DifficultyBeginner
At a glance
- Visits from ChatGPT, Perplexity, Gemini and Copilot show up under their own domains when the referrer survives.
- Referrers get lost in desktop apps, in-app browsers and copied links, so the visible count undercounts the real one.
- Assistant referrals tend to land deep, on the exact page that was cited, rather than on a homepage.
Where an assistant referral lands in your reports
A visitor clicks a citation in a Perplexity answer. Their browser sends a referrer header naming perplexity.ai, your analytics reads it, and the session files under referral.
That is the clean path. Several things break it. Native desktop apps often pass no referrer. A link opened in an in-app browser may report the app instead. A user who copies the URL into a new tab sends nothing.
Everything that breaks arrives as direct, the same bucket as bookmarks, typed URLs and untagged email. You cannot separate them afterwards, and that silent portion is dark traffic. Server logs catch a little more, because log file analysis sees requests that never fire a tracking script.
Small numbers that behave nothing like search traffic
Two teams read the same report opposite ways. One sees a rounding error and stops. The other checks what those sessions did.
Assistant referrals arrive further along than a search visit. The buyer has read a summary, compared options and chosen which link to follow, so they land on a specific page with the question mostly answered.
Judging that on volume against organic search will always disappoint. Judging it on what happens after the landing, and on the AI-attributed conversions that follow days later, is the fairer read. The visible traffic is not the prize yet. It is the receipt for visibility you earned somewhere you cannot see.
Split assistant referrals out of direct this week
- Create a channel group for assistant domains. List chatgpt.com, perplexity.ai, gemini.google.com and copilot.microsoft.com so they stop scattering across referral and other.
- Set a direct-traffic baseline now. Record the current level per landing page before any GEO work. That comparison is your only lever on the invisible share.
- Look at landing pages, not the homepage. If one deep page takes the bulk of assistant referrals, read it and work out why it got cited.
- Compare behavior, not volume. Pages per session, scroll depth, add to cart. Volume will look bad for a while.
AI search traffic vs organic search traffic
| Organic search | AI search | |
|---|---|---|
| Referrer | Reliable, engine named | Often missing, lands in direct |
| Landing page | Mix of homepage and deep pages | Mostly the specific page cited |
| Visitor state | Still comparing options | Already shortlisted you |
| What the volume means | A proxy for demand | A fraction of the exposure you got |
| Best paired with | Rank and click data | Prompt appearance and citation data |
Frequently asked questions
-
Is AI search traffic the same as referral traffic?
Partly. The share that keeps its referrer does sit in referral, under the assistant’s domain. The rest arrives as direct because the referrer was stripped in an app or a copied link. Treating your referral figure as the whole of AI search traffic will understate it, sometimes badly.
-
How do I see AI search traffic in Google Analytics?
Build a custom channel group matching the assistant hostnames and apply it to new and historical data where your setup allows. That gets you the visible share in an afternoon. The invisible share has to be inferred from direct-traffic movement against a baseline you recorded earlier.
-
Is it worth chasing when the volume is this low?
Not as a revenue channel this quarter. The traffic on its own will not hit a number, and saying otherwise sells you something. What it is worth is the behavior of the visits that do land and the baseline you build before the volume arrives. Start cheap, keep measuring, do not promise sessions.
-
Will this traffic grow into something like search volume?
Nobody knows, and anyone handing you a figure is guessing. What is observable is that assistant usage keeps rising while click-through from answers stays low by design, since the answer is meant to satisfy the question. Plan for influence without clicks, not for a second organic channel.