Skip to content

Introducing AI Traffic Tracking and Conversion Attribution

Ali Khallad4 min readUpdated
April 12, 2026 , 4 min read
SurfacedBy product update hero: headline "See which AI platforms visit, cite, and convert" beside a screenshot of the Tracking page install methods.
Share

A visibility score tells you an AI assistant named your brand. It does not tell you whether anyone followed the answer to your site, whether the assistant’s crawler ever reached your pages, or whether any of it turned into a sale. AI traffic analytics in SurfacedBy closes that gap by joining three things we used to look at separately: the visits AI assistants send you, the crawlers that fetch your pages, and the conversions those visits produce.

What launched

The AI Traffic view now brings four signals into one place:

  • AI-referred sessions from your connected analytics, so you can see visits that arrived from ChatGPT, Perplexity, Google AI Mode, and other assistants.
  • Conversions and revenue on those sessions, wherever the connected property reports them.
  • Bot and crawler access from our on-site tracker, so you can see which AI crawlers fetched which pages.
  • Install status by collection method, so you always know what is being measured and what is not.
SurfacedBy AI traffic metrics: 326 AI visitors, 18 AI conversions, $4,240 attributed revenue, and 4,512 AI bot crawls, each with a 30-day change.
One row ties AI visits to conversions, attributed revenue, and crawler volume over the last 30 days.

Why we built it

Citation tracking answers one question well: did an AI answer mention you. It says nothing about what happened next. A brand can be cited often and see no visits, or get steady AI traffic that never converts, or have its pages fetched by a crawler whose answer never cites it. Those are different problems with different fixes, and you cannot tell them apart from a visibility score alone. Tying citations to traffic, crawl access, and conversions is what turns “we are mentioned” into “here is what the mention is worth.”

The line between a crawler fetching your page and a person clicking through to read it is not something we invented for the product. We worked it out first in a probe of real nginx logs, watching which assistants fetch live, which only send human clicks, and which are invisible either way. This feature is that research turned into something you can measure without grepping logs yourself.

What SurfacedBy tracks

SignalSourceWhat it tells you
AI-referred sessionConnected analyticsA person arrived from an AI assistant’s link
Confirmed visitOn-site tracker or pluginA real human hit, including cases the analytics referrer misses
Crawler fetchTracker collectorsAn AI bot fetched the page, labeled by what it was doing: answering live, indexing for search, or collecting for training
Conversion or revenueConnected analytics propertyA session that completed a goal or purchase, where the property reports it
SurfacedBy Tracking card splitting AI bot activity into on-demand retrieval (412), search indexing (1,120), and model training (2,980), each with top bots and an upward trend.
On-demand retrieval, search indexing, and model training, separated by bot. Crawler fetches are counted apart from human visits.

Where the numbers stop

Measuring AI traffic honestly means being clear about its limits, so this view is built to under-claim rather than over-claim.

  • Correlation stays labeled as correlation. When citations rise alongside AI-referred visits, that is a relationship worth watching, not proof that one caused the other.
  • We do not pretend to see every AI visit. Some AI traffic arrives with no identifiable referrer, so we report the AI traffic we can confirm and flag the gaps instead of inflating one number. Google is the clearest case: its assistants ground on the same index Googlebot builds, so a live Gemini fetch is not separable from ordinary search crawling.
  • Crawler fetches and human visits stay in separate counts. A bot fetching a page is not a person reading it, and folding the two together would make both useless.

How to turn it on

  1. Connect your analytics in Settings, then pick the property you want to read from.
  2. Install the on-site tracker or the plugin so confirmed visits and crawler fetches start flowing.
  3. Open the Tracking page and the AI Impact card on your Overview to watch visits, crawler access, and conversions next to your visibility.

Connected analytics alone gives you AI-referred sessions and conversions. Adding the tracker is what surfaces crawler access and confirms visits that analytics cannot attribute on its own. The two are designed to layer.

Where to look first

If you already track a domain with us, connect a Google Analytics property and watch the AI Impact card for a week. The first thing most teams notice is the distance between how often they are cited and how often that citation becomes a visit. That distance is the reason this view exists. If you want to see the raw signal behind it first, our nginx log breakdown of AI traffic versus referral traffic shows exactly what each assistant sends when it reaches your origin.

A note on timing: AI traffic tracking changes often. Some of that is customer requests, some is what we learn from our own research and testing, and a lot is just keeping pace as new AI crawlers and platforms appear. So treat the numbers and screenshots here as how the feature looked at launch. If your dashboard shows more than what is here, you are on a newer version.