"Marketing is just deadweight," I overheard a CFO say at a conference this summer.

How sacrilegious. Isn't global ad spend supposed to grow 7.8% and nearly top $1 trillion this year? GroupM has just revised its estimates for the year upwards and now expects digital to grow 10%, out-of-home 12%, retail media 18%, and CTV a whopping 20%. Can so much money be all wrong?

I almost dismissed that gripe, but then recalled a recent note from Gartner that companies today are facing a deadweight economy (that D-word again). "Growing economic optimism in advanced economies obscures an inconvenient truth: favorable conditions that have powered growth in the last decade are no longer present," the research firm pointed out, before warning CFOs against cooling demand, increased price sensitivity from indebted consumers, productivity stagnation, and a litany of other challenges.

These two waves — the expectation that ad spend will continue to increase indefinitely, and growing caution from those holding the purse strings — are on a collision course. If ever there was a time for marketing to prove its worth, it's now.

Keep things simple

What tools should marketers use today to demonstrate the return on their marketing investments?

At InfoSum, we’re big believers in the power, rigor and simplicity of incrementality testing. Legacy measurement techniques like multi-touch attribution (MTA) and marketing mix modeling (MMM) require massive amounts of data to fuel complex formulas that are costly and time-consuming to execute. Designed to bring some clarity to multi-channel measurement, they’re not particularly effective at handling fast-changing market conditions, intense channel fragmentation, or the realities of data sharing (or lack thereof) in a world dominated by walled gardens.

Incrementality tests, on the other hand, can be done in a matter of weeks, zero in on very specific tactical questions, and provide answers that can be immediately rolled out to improve any campaign.

What is incrementality testing?

You’re already familiar with the principles behind incrementality testing. Think back to your science experiments in middle school: perhaps you grew green beans under different lighting conditions, or you studied how fast planaria regenerate at different temperatures. You split your subjects randomly into test and control groups and compared their outcomes.

It’s the same thing with incrementality (or lift) experiments in marketing: The test group (A) is exposed to the campaign, the control group (B) is not, and since both groups are assigned at random (and thus equally likely to be influenced by outside factors), the difference you see in post-exposure behavior between A and B  — be it measured in sales, website visits, app downloads or any other key outcome metric for the brand — can be safely attributed to the campaign.

That’s a crucial benefit of incrementality testing over other measurement methods: it captures causality, not just correlation between ad exposure and outcome.

Return to the future of marketing

To be clear, incrementality testing isn’t a new thing, even in marketing.

Early marketers fully understood the need for scientific measurement. As Daniel Starch, one of the fathers of market research, put it a hundred years ago, “the judgment even of expert advertising men is often wrong, and they are hardly better judges of what will appeal to the public at large than other observers of human nature. The fact that so many weak advertisements appear even in our best mediums and for products for which large sums of money are expended is ample proof of this statement.”

If there’s a better justification for the use of scientific methods in marketing, I’m not aware of it. Starch went on to publish dozens of fascinating ad copy experiments in Principles of Advertising for iconic brands like Colgate, Campbell’s, Cadillac, Cream of Wheat, Beech-Nut, Kodak and many others. Even in 1923, those experiments were conducted with a keen understanding of statistical sampling, test and control selection, and population representation.

New requirements

While modern tests and experiments have a lot in common with the tests of yore, there are important new steps to consider.

One has to do with the selection of test participants. Today’s advertising ecosystem is highly complex, and it can be difficult to figure out who is exposed to what and on what platform. For randomization to do its trick, market researchers need to make sure that they’re starting from a reasonably uniform base. If the subjects in the control group happen to like a competitor brand, use more social media, or watch different channels, results will be skewed. Marketers need to watch out that the way they plan to execute a test, and the ad tech partners they pick for the task, don’t inadvertently introduce selection bias.

The second consideration has to do with data privacy. That’s a big one. To run a lift experiment today, brands and their media partners need to stay away from intrusive tracking techniques. It doesn’t matter that cookies are still around, marketers need to invest in solutions that respect the privacy of their users. The solution is to bring together separate transaction and exposure data and make sure that the partners can share in the insights without divulging their respective first-party data to one another.

Sharing without sharing? That’s where a strong data clean room can make a big difference.

The data clean room difference

Our data clean room philosophy at InfoSum is the non-movement of data. We’ve developed advanced data storage, user access and analysis techniques that make it possible for data partners to connect proprietary data sources without ever sharing a single piece of data. We’ve built a comprehensive data collaboration platform on that foundation, and brands and media owners use it today to conduct tests and experiments without having to worry about the thorny legal, operational and competitive complications that come with traditional data sharing agreements.

This is how Deliveroo and Channel 4 were able to unequivocally attribute a smashing 20% lift in app signups to a new advanced TV campaign, and how Renault and Axel Springer All Media were able to accurately compare the performance of a first-party campaign against a traditional cookie-based campaign.

Thanks to broad relationships with global data and tech companies like WPP’s Choreograph, global media agencies like Canvas Worldwide, and media heavyweights like Netflix and News Corp Australia, many more brands across industries now have a chance to use InfoSum’s platform to unlock their testing capabilities.

Make an impact today

I’m not dismissing MMM and MTA entirely. There’s still a time and place for those measurement solutions. But brands are starting to move away from open RTB and opt for more direct integrations and private marketplaces. When 80% or 90% of a brand’s media budget goes to a single media partner — a social media platform, for instance, or a retail media network, or a large connected TV player — multi-channel measurement isn’t such a big concern anymore.

Incrementality has speed, agility, cost, privacy and scientific integrity going for it. You can do a lot worse to get your CFO to back off the deadweight insinuations and become your biggest supporter.