METRICS

AI Referral Traffic

An isolated narrow channel carries a faint luminous flow, distinctly separate from broader surrounding dark currents, symbolizing AI referral traffic as a unique independent channel.
AI referral traffic, though small in proportion, directly proves AI visibility converts to visits and must be tracked as a separate channel.

IN ONE SENTENCE

AI referral traffic is still small in absolute terms, but it is the clearest evidence that AI visibility turned into visits — so it needs its own channel rather than being folded into direct traffic.

AI referrals are still a small share of total site traffic, but they are growing quickly and they are among the few pieces of direct evidence that AI visibility produced a business action. The problem is that default analytics configurations bury them in Direct / Other, where they disappear.

OUR POSITION

Do not present this metric as a multiple. Off a tiny base, multiples look impressive and persuade no one. Absolute numbers plus its share of enquiries is the version that holds up.

01

Identifying it

Create a dedicated channel group by referrer domain so the major AI platforms are recognised instead of landing in Direct / Other.

Some platforms pass no referrer at all. That portion can only be estimated from landing-page patterns and timing correlation — and estimates must be labelled as estimates.

Do not forget the indirect path: people who see the brand in an AI answer and then search the brand name or type the URL carry no AI referrer, but they are still a product of AI visibility.

02

What the volumes look like

Cross-site studies put the average at roughly one percent of traffic. In absolute terms this remains orders of magnitude below conventional search.

Growth is a different story, with referral volume multiplying within a single year. The value of this metric is in trend and structure, not in this quarter's volume.

03

On the claim that AI traffic converts better

Several studies point the same direction but disagree wildly on magnitude, from a few times to more than ten, using very different methods.

There is also counter-evidence: at least one statistical analysis found no significant overall difference. The defensible phrasing is that it is worth watching separately — not that it converts some specific multiple better.

Data behind this page

1.08%

AI referrals as a share of total site traffic

SourceConductor, across 13,770 domains,2025

1.13B

Monthly AI-platform referrals to the top 1,000 sites

SourceSimilarweb, via TechCrunch; up 357% YoY,2025-06

8%

Click-through on a traditional result when an AI summary is shown

SourcePew Research Center, 900 US adults / 68,879 searches (15% without a summary),2025-07

Sources

  1. [1]Google users are less likely to click on links when an AI summary appears.Pew Research Center.2025-07-22

Updated 2026-08-10