Marketing Attribution Is Breaking: What CMOs Should Measure Instead

By the IAB's State of Data 2026, 75% of buy-side leaders say the core methods they rely on — attribution, incrementality testing and marketing mix modeling — are underperforming, and signal loss has erased an estimated 30–40% of trackable conversions. The reflexive conclusion is that attribution is dead. That is the wrong lesson, and acting on it produces worse decisions, not better ones.

Attribution is breaking as a reliable record of touchpoints. But most CMOs were asking attribution to answer a question it was never built for. The useful move in 2026 is not to replace one attribution model with another. It is to separate two questions the industry has spent a decade conflating.

Attribution and measurement are not the same question

Attribution asks: which touchpoint should get credit for this conversion? Measurement asks: did marketing cause an incremental business result that would not have happened otherwise? These are different questions with different methods, and a decade of dashboards taught marketers to treat the first as if it answered the second.

It never did. A last-touch or multi-touch model can assign 100% of credit across channels and still tell you nothing about incrementality — whether the customer would have converted anyway. When the touchpoint record was cheap and complete, the confusion was survivable. Now that the record is degrading, it is expensive: a CMO optimizing budget against attribution is optimizing against a measurement that was never causal and is now also incomplete.

Why the touchpoint record is degrading — and it is not just cookies

The privacy story is real but partial. Apple's App Tracking Transparency, consent regimes and walled gardens have degraded user-level tracking, and the trackable-conversion loss the IAB estimates at 30–40% is the visible result. But treating this as a cookie problem understates it.

The deeper shift is that the buying journey is moving off the click entirely. Zero-click search and AI-generated answers mean buyers form shortlists without a trackable visit; AI-mediated and agentic buying compresses the funnel into interactions no pixel sees. Open Future Forum's marketing research captures how fast this is arriving — 81% of marketing and growth respondents are past exploring agentic AI, and as agentic buying changes the funnel, attribution is being re-litigated, not refined. The record is not just noisier. Parts of the journey have stopped producing a record at all.

MMM is back, and what it is good for

The privacy-era answer marketers are converging on is an old one: marketing mix modeling. MMM is econometric, not user-level — it infers the contribution of channels from aggregate spend and outcomes, so it needs no cookies, device IDs or identity graph. Open-source tools from Google (Meridian) and Meta (Robyn) have collapsed the cost of entry, and roughly half of US marketers now plan to invest in it.

MMM is not a silver bullet. It is coarse, it needs history, and it struggles with fast-changing channels. Its value is that it answers a question attribution cannot — the incremental contribution of spend at the portfolio level — without depending on a per-user record that is disappearing. The right framing is MMM and attribution answering different layers of the same problem, not one replacing the other.

The measurement hierarchy CMOs should run

The practical rebuild is a hierarchy, top-down, where each layer answers a question the layer below cannot.

• 01Business outcome — did revenue, pipeline or contribution margin move? The number the board and CFO actually care about.
• 02Incremental contribution — did marketing cause it? Answered with incrementality experiments and MMM, not attribution.
• 03Channel allocation — how should budget split across channels? MMM's home ground.
• 04Campaign optimization — which creatives and campaigns to adjust? Attribution and platform data are fit for this.
• 05User-level attribution — used only where the signal is still reliable, and understood as an optimization tool, not a truth.

The IAB's own 2026 guidance lands in the same place: triangulation — MMM plus incrementality testing plus attribution — rather than any single source of truth. The discipline is to stop letting a lower layer answer a higher question: never let campaign-level attribution stand in for whether marketing caused the business result.

The CMO now owns the seat where this converges

There is a reason this lands on the CMO specifically. The CMO already owns attribution and marketing measurement, and in 2026 the CMO is also becoming the AI buyer for the function — in Open Future Forum's research the marketing AI buying decision is converging on the seat that owns attribution, with most marketing buyers already piloting or running agents in production. That makes the measurement question and the AI question the same person's problem: the tools reshaping the funnel and the tools measuring it now report to one seat.

So the CMO's task is neither to defend attribution nor to declare it dead, but to run the hierarchy — measure incrementality at the top, use attribution as a lower-level optimizer, and stop moving budget on the strength of a touchpoint record that is quietly disappearing.

Last updated: August 19, 2026

Murray Newlands

Founder, Open Future Forum

Murray Newlands has been building executive communities in Silicon Valley since 2019. Open Future Forum runs role-specific forums and curated gatherings for senior executives and investors, grounded in a give-first philosophy.

About Murray LinkedIn Twitter / X Substack