The measurement problem everyone pretends to solve

At a conference in Miami last month, the director of e-commerce strategy at Nestlé said something that should have made headlines.

She said that marketers have been trying to measure incrementality for decades.

And there just aren't great solutions.

This is Nestlé. One of the largest advertisers on the planet. Hundreds of analysts. Budgets that most companies cannot imagine. And they stood on a stage at one of the industry's most prestigious events and said: we don't know if our campaigns work.

Neither does Haleon. Neither does Molson Coors.

I was not surprised.

The promise that built an industry

For two decades, digital marketing sold one idea above everything else.

Unlike television. Unlike print. Unlike billboards. Digital is measurable.

You will know exactly which ad drove which click. Which click became which lead. Which lead became which customer. A perfect chain of attribution, visible in your dashboard, updated in real time.

It was a compelling promise. It was also, in large part, a fiction.

Not a deliberate one. The technology genuinely seemed to deliver what it claimed. Click tracking worked. Conversion pixels fired. Reports looked precise.

The problem was structural, and it was there from the beginning.

The judge and the jury

When Meta tells you your Meta campaigns are working, Meta is measuring Meta's performance using Meta's data inside Meta's platform.

When Google tells you your Google campaigns are working, Google is the measurement provider for Google's own media.

Every major platform is simultaneously the seller of advertising and the auditor of its own results. This is not a conspiracy. It is a business model. And it creates a systematic incentive to present results favourably - not through fraud, but through the quiet choices of what to count, what attribution window to use, which conversions to claim.

Two platforms claiming credit for the same conversion is not an anomaly. It is the default.

I watched this happen for years before it had a name. I watched this happen for years before it had a name. Some years ago, my team was brought in to audit the full marketing operation of a beauty brand - advertising effectiveness, strategy, UX, growth opportunities. The kind of engagement where the brief is broad because nobody is quite sure where the problem is.

We found it quickly.

A beauty brand I worked with was spending around 15,000 euros a month on Google Ads.

The agency reported thousands of conversions every month. The numbers looked excellent.

They were counting brand searches.

The brand had a complex, three-part French name. Existing customers could not remember the exact spelling. So they typed it into Google. The ad appeared. They clicked. The agency counted a conversion.

These were not new customers. They were loyal customers who would have found the brand anyway - the organic result was sitting directly below the ad, in first position. The brand was paying to intercept its own existing audience, month after month, year after year. Less than five percent of the budget was reaching anyone new.

The business was growing regardless - carried by physical locations in shopping centres. So nobody looked closely at the numbers. The agency had no reason to flag it. The dashboard showed green.

This is not a story about a bad agency. It is a story about a measurement architecture that made the problem invisible.

What actually broke - and when

Rand Fishkin documented the mechanism clearly in 2024, with data that is worth reading if you have not seen it yet.

Clicks are dying. Zero-click consumption - where people read, watch and decide without ever leaving the platform - has fundamentally changed how influence works. Social platforms algorithmically reward content that keeps users inside. So content stays inside. Brands build audiences they can no longer reach without paying. Influence happens somewhere that cannot be tracked.

At the same time, the tracking infrastructure that digital marketing was built on is collapsing. Apple's privacy changes removed a significant portion of third-party data from Safari. Ad blockers now reach between a third and half of all internet users - and they block tracking, not just ads. Privacy regulation has restricted what can be collected and how long it can be kept. Multi-device journeys make identity resolution increasingly approximate.

The result is a measurement gap that widens every year.

And into that gap, platforms report confidently. Because their incentive is to report confidently.

What I do instead

I want to be direct about something.

I do not fix attribution. Nobody does. Not with the tools available to companies spending less than several million a year on measurement infrastructure alone.

What I do is help founders and CMOs make better decisions despite broken attribution. That means identifying which numbers are real and which are platform narrative. Revenue is real. Customer retention rate is real. Marketing Efficiency Ratio - total revenue divided by total spend - is real. These are numbers that no platform dashboard can manufacture, because they come from your bank account, not from a pixel.

It means designing a measurement architecture where no single platform's self-reported numbers can move a budget unilaterally. Where every channel's claimed performance has to survive contact with revenue reality before it changes anything.

And it means being honest with clients about what cannot be known - and building strategy around that honesty rather than around comfortable fictions.

The brands that will win the next decade are not the ones with the most sophisticated attribution model. They are the ones that stopped pretending the model was telling the truth.

Robert Sosnowski is the founder of Growth Architects Studio. He has worked with 200+ brands across Europe and the US over 22 years in brand and growth strategy.

Rand Fishkin - The Death of Click-Based Attribution (2024) SparkToro founder presents data showing the collapse of click-driven measurement and the rise of zero-click consumption across all major platforms. https://www.linkedin.com/feed/update/urn:li:activity:7221294276238852096/

Adweek - Brands Like Nestlé, Haleon, and Molson Coors Wrestle With Broken Measurement (April 2026) A report from the Possible conference in Miami where senior marketing leaders from global brands publicly acknowledged that measuring incrementality remains an unsolved problem - decades in the making. https://www.adweek.com/media/what-to-expect-from-possible-2026/

eMarketer - FAQ on Incrementality: How to Prove Your Ads Actually Work (2026) An overview of measurement methodologies - MMM, multi-touch attribution, and incrementality testing - with data showing that only 27.6% of marketers consider MMM the most reliable method, and that barriers around accuracy, tools, and cross-platform application persist. https://www.emarketer.com/content/faq-on-incrementality-how-prove-your-ads-actually-work-2026