The invisible influence problem
Traffic from AI search does not show up in analytics the way referral traffic does. A user gets an answer in ChatGPT, Perplexity or Claude, copies the product name and arrives via direct entry or a Google search. The referrer is lost. Analytics logs it as direct or organic, though the real source was AI.
That creates a blind spot. You cannot see where leads come from or whether your presence in AI results is working.
Attribution lies
Last-click attribution records the final source before conversion. But the user journey from AI search works differently: they read a recommendation in a chat interface, remember the brand, then type the name into a browser or search it through Google.
The analytics system credits the conversion to direct or branded search. AI stays offstage, even though it started the chain.
Referrers from AI platforms are unreliable too. Some clicks get masked, others are lost when a link is copied. Relying only on UTM tags and referral data means undervaluing the channel.
How to measure real influence
Registration surveys. Add a "How did you hear about us?" field with an option that explicitly names AI: ChatGPT, Perplexity, Claude, Gemini. Do not bury it in a generic "search engine" — users distinguish between sources and will answer honestly.
Direct comparison with known channels. If surveys show AI brings more signups than YouTube but fewer than Google, that is a concrete metric for prioritisation. Numbers from registration forms are more accurate than guesses based on referral traffic.
Brand mention tracking. Set up monitoring for branded search queries: growth in direct visits for your product name may correlate with appearing in AI results. If people started searching for your brand more often after ChatGPT began recommending your service, the connection is clear.
A/B testing through search presence. Optimise content for specific queries, track whether you appear in AI model answers, record changes in traffic. If direct and branded organic grew after optimisation, the AI channel likely worked.
Why this matters now
The share of search queries going through AI is growing. Ignoring a channel that already beats YouTube on leads means losing conversions.
Most companies still do not isolate AI as a separate source in analytics. That gives an edge to those who start measuring early: you will see which pages and topics AI models pick up and can strengthen your presence before competitors understand what is happening.
What to do right now
Add a source question to your registration form or onboarding flow. It takes an hour of developer time but delivers data that Google Analytics does not have.
Check whether ChatGPT, Perplexity or Claude mention you for key queries in your niche. If not, optimise your content: structured answers, clear definitions, tool lists. AI models prefer content that is easy to quote.
Set up a weekly report on branded search and direct traffic. If those metrics grow without visible changes in advertising or PR, check AI results. You may already be getting recommended without knowing it.
