How Marketing Measurement Is Changing with Prof. Koen Pauwels

For years, conversations about incrementality, marketing mix modeling (MMM), and causality lived mostly in research papers and analytics teams. Today they’re boardroom topics. CMOs are under pressure to prove marketing’s contribution to business growth, CFOs want evidence instead of assumptions, and marketers are realizing that platform-reported ROAS isn’t the same as incremental impact.

That’s exactly why we launched Humans of Measurement, a new segment of Lifesight’s Profit & Proof podcast featuring the people shaping the future of marketing measurement.

For our first episode, I spoke with Prof. Koen Pauwels, Distinguished Professor and Associate Dean of Research at Northeastern University, former Principal Scientist at Amazon Ads, and one of the world’s leading experts on marketing effectiveness.

Our conversation covered everything from MMM and experimentation to AI and attribution. Here are the biggest takeaways.

Why Causal-Measurement Is Taking Center Stage

Marketing measurement didn’t suddenly become important because third-party cookies disappeared. According to Koen, three larger forces are driving today’s shift.

First, marketers are moving beyond deterministic attribution because it was never designed to answer strategic business questions. As Koen put it:

“If I spend 20% more or less on a certain channel, will I get incrementally more profit or not?”

That’s ultimately the question executives care about, not whether one individual clicked an ad.

Second, attribution has always left gaps. Television, out-of-home advertising, word of mouth, and countless offline interactions have never been fully measurable through user-level tracking.

Finally, finance has entered the conversation. As CFOs began comparing reported ROAS with actual business performance, they started asking a simple question:

“If marketing is performing so well, why isn’t the business growing?”

That question has accelerated the industry’s shift toward incrementality and causal measurement.

Open-Source MMM Made Measurement More Accessible, Not Automatic

Many marketers credit open-source tools like Robyn and Meridian for the resurgence of MMM. Koen sees it differently.

The real breakthrough wasn’t better models. It was better access to data and easier implementation. As he explained, most MMM projects historically spent the majority of their effort preparing data, not building models. Open-source frameworks lowered the barrier to entry, allowing smaller marketing teams to start benefiting from MMM much sooner.

But easier software doesn’t eliminate the need for strategic thinking. One of Koen’s most memorable observations perfectly captures the challenge facing modern marketers:

“If you’re not willing to change any decision, then my ROI is zero.”

Measurement creates value only when organizations use it to make better decisions.

Start With Decisions, Not Data

One theme surfaced repeatedly throughout our discussion. Too many organizations begin with available data instead of business decisions.

Before building a model, Koen encourages leadership teams to map the customer journey and identify the business questions they actually need answered. Only then should they decide which metrics belong in the model.

This philosophy closely aligns with Lifesight’s approach to unified marketing measurement. Models, experimentation, and attribution each answer different questions, but together they create a decision framework marketers can trust.

Why Every Company Needs Its Own Measurement Language

Marketing terminology is surprisingly inconsistent. Concepts like adstock, halo effects, long-term impact, and return on marketing investment often mean different things to different organizations.

Koen believes the solution isn’t industry-wide perfection. It’s internal alignment.

Every stakeholder, from marketing to finance, should share the same definitions so decisions are made from a common understanding of performance. Consistency inside the business matters far more than universal agreement across the industry.

Experimentation Is Still the Gold Standard

Most marketers agree experimentation is important. Far fewer actually run meaningful experiments. According to Koen, technology is no longer the obstacle. Risk is.

Organizations hesitate to pause campaigns or reduce spend because the perceived downside feels too high. His solution is an iterative framework he calls MEME:

• Model
  
• Experiment
  
• Model
  
• Experiment

Historical models generate hypotheses. Experiments validate those hypotheses. The experimental results improve the next generation of models. The cycle repeats, creating increasingly reliable measurement over time.

At Lifesight, we often recommend a similar approach using geographic holdout tests to validate model predictions and establish a true incremental baseline.

Attribution Still Matters, Just Not for Everything

One of the most important distinctions Koen made was between reliability and validity.

Advertising platforms are generally reliable. They consistently report clicks, impressions, and conversions. But they’re not designed to measure incrementality. Platforms only see what happens inside their own ecosystem. They don’t know what happened across television, retail media, search, Amazon, or every other marketing channel influencing customer behavior.

That’s why attribution still has an important role, primarily as an operational optimization tool rather than a complete measurement system. Strategic investment decisions require broader measurement approaches like MMM and experimentation.

Where AI Helps, and Where Human Judgment Still Wins

AI is rapidly changing marketing measurement. But Koen cautioned against confusing automation with expertise.

Experienced practitioners often become dramatically more productive using AI. Less experienced teams, however, may spend more time correcting AI-generated mistakes than solving real business problems.

His warning about “cognitive surrender” resonated throughout the conversation. AI should accelerate human expertise, not replace critical thinking. The organizations that succeed won’t be those using the most AI. They’ll be the ones using AI to strengthen better decisions.

Better Measurement Starts With Better Decisions

Perhaps the strongest insight from our conversation wasn’t about algorithms, attribution models, or AI. It was about organizational behavior.

Marketing measurement succeeds when people act on it. As Koen put it:

“Working backwards from the decision is always key.”

That’s ultimately the goal of modern measurement, not building increasingly sophisticated dashboards, but giving leaders the confidence to make better business decisions.

As marketers continue combining marketing mix modeling, incrementality testing, experimentation, and attribution into unified measurement frameworks, the organizations that win won’t necessarily have the most data. They’ll be the ones that make the best decisions with it.

Watch the Full Conversation

 

This article highlights just a few insights from our discussion with Prof. Koen Pauwels on the Humans of Measurement podcast.

Watch the full episode to hear our conversation on marketing mix modeling, experimentation, incrementality, attribution, AI, and the future of marketing measurement.  Watch on YouTube, listen on Spotify, or listen on Apple Podcasts.

Ready to build a unified measurement strategy? Learn how Lifesight combines MMM, incrementality testing, experimentation, and attribution into a single decision framework that helps marketers measure what matters and invest with confidence. Book a Demo