Your Attribution Model Is Lying to You in the AI Era.

Somebody asked ChatGPT which CRM to use for a small agency. ChatGPT mentioned your product by name, described exactly why it fit their use case, and the person then opened a new tab, typed your brand name directly into Google, clicked your ad, and converted. Your dashboard now says “Paid Search, Branded Term” got the credit.

That’s not what happened. That’s what your attribution model is capable of measuring. Those are two very different things, and the gap between them is getting wider every month.

The Original Sin

Last-click was always a lie, we just tolerated it

Marketers have known for years that last-click attribution overweights bottom-funnel channels and underweights everything that built awareness earlier in the journey. We tolerated it because it was simple, universally understood, and good enough when most discovery still happened inside channels we could actually track with a pixel.

AI-mediated discovery breaks that tolerance completely. When someone gets a recommendation inside a private AI chat session, there’s no referral URL, no UTM parameter, no cookie trail. The conversation that actually convinced them happens in a black box your analytics stack has zero visibility into. Then they show up through a channel you can track, and that trackable channel gets 100% of the credit for a decision that was mostly made somewhere else entirely.

Why This Matters More Than It Sounds

You’re about to defund the thing that’s actually working

Here’s the dangerous part. If AI citations are quietly doing real work influencing buyers, but your dashboard can’t see it, the natural move is to cut the budget and effort going toward earning those citations, because it “isn’t showing results.” You’d be defunding your best-performing, least-measured channel because your measurement stack wasn’t built for it. That’s not a hypothetical. I’d bet money it’s already happening inside marketing teams right now, and nobody’s noticed yet because the reporting looks fine.

The Actual Proposal

A citation share plus assisted conversion stack

I’m not going to pretend there’s a clean, off-the-shelf tool for this yet. There isn’t. But here’s the measurement stack I think teams need to start building toward:

• Citation share: manually and systematically track how often your brand gets mentioned when you (and ideally a sample of real customers) ask AI tools questions relevant to your category. Log it monthly like you would a rank tracker.
• Branded search lift: watch for unexplained increases in branded search or direct traffic that don’t correlate with a specific campaign. That’s often AI-influenced discovery surfacing downstream.
• Post-purchase attribution surveys: a simple “how did you hear about us” question at signup or checkout will catch AI-influenced journeys that no pixel ever could. This is the least glamorous fix and the most reliable one.
• Assisted conversion, not last-click, as the primary internal metric for anything content or brand related. Stop rewarding the channel that closed the deal and start crediting the channels that opened it.

Quick Answers

A few direct questions, answered directly

Why is last-click attribution becoming less accurate?
Because a growing share of buyer discovery now happens inside AI chat tools that leave no trackable digital trail, so credit defaults to whatever trackable channel closes the conversion, even if it wasn’t the actual influence.

How do you measure AI-influenced conversions?
There’s no mature automated tool yet. The most reliable current methods are manual citation tracking, watching branded search and direct traffic trends, and post-purchase “how did you hear about us” surveys.

What is citation share in marketing?
An emerging, mostly manually tracked metric describing how often a brand is mentioned or recommended when relevant questions are asked inside AI tools like ChatGPT, Perplexity, or Google AI Mode.