In the last 5 years, the number of media channels available to marketers has increased by at least a factor of two or three. We’ve been so busy dealing with the change, that we haven't really looked into whether our current methods for measurement and optimization of media spend can handle this data explosion. So we need to ask ourselves . . . Are we doing okay?"

AdTech has evolved media options that attempt to reach into all aspects of consumer behavior. There are now more categories of media, like social media, retail media, and streaming. They appear on various screens, and there are hundreds of vendors operating in each separate digital space. Each type of media has also developed different ways to target specific audiences and screens. For example, social media has different tiers of influencers, while some media like digital out-of-home and Waze aim to target local mobile devices.

From the IAB Retail Media Buyer’s Guide, July 2023, Source: eMarketer, Oct 2022

Parallel with the growth of AdTech, and often driving it, there are also more places for consumers to make purchases. Many consumer packaged goods companies now have their own direct-to-consumer sales channels. Giants like Amazon and Walmart have evolved from places to buy things to advertisers in their own right. The growing complexity of media is now matched by a complexity in sales channels and a blurring of roles. Which ads are driving which sales? Marketing mix models (MMM) have traditionally been good at capturing this, but now they're facing daily challenges.

Multi-touch attribution (MTA) emerged briefly as the big solution to all this complexity.  However, privacy concerns and walled gardens have limited the potential of MTA, pushing it into more controlled environments with limited single-vendor data. The task of integrating all marketing activities has now fallen back on marketing mix models. But can MMM handle these new challenges? Has it evolved to keep pace with the media it tries to measure? Nothing in this world can be stretched and not break if pushed too far. Where are we pushing MMM and what should buyers be concerned about?

We need to start thinking of MMM as one tool (a big one) in a marketing measurement framework that includes agile testing, brand lift measurements and other trackers.  Instead of trying to include more and more media channels (and sub-channels) in the MMM, we should decide what are the big questions we want it to solve? The main task of the MMM is to provide a holistic view. There are lots of alternative approaches to get to details. We’ve been working under the assumption that we can add new media and promotion variables to the model at any time. Now we have to consider what's essential for an integrated view of effectiveness and what's not. How can we simplify the models? This thinking should extend also to how far we should go in applying the various secondary models (often called “deep dives” or “outboard models”) that are used to break down media into format, content, and partner levels. 

Much of the promise of MTA was that it would reach down into the ultimate level of granularity – the individual, in a particular phase of the path the purchase. AdTech was already going there by targeting specific audiences, monitoring search, and creating interactive environments on social media and YouTube. MMM can do a much better job of aligning itself with the granularity and context of the media. In general, MMM should try to get down to the level of granularity at which the consumer sales response is seen. Sometimes this is difficult because different depths and quality of data are available for each media and promotion channel. For instance, you may have access to Google data through Ads Data Hub and you may have a vendor that has store-level data but can either be tied to Meta data below the DMA level? Can Google data be joined to Meta data? Usually not. MMM is the natural tool to bridge these divides but there are many different ways to join up the data, and modelers are only now coming to grips with what works better.

Another key trend driving the marketing measurement industry is the need for speed. Since AB testing is such a dynamic and forward-looking enterprise, marketers expect the same thing from MMM. Using MMM is better than seat-of-the-pants solutions but not if it comes months late. How can MMM be less backward-looking? One way is to focus on new ways to optimize and make predictions based on combining knowledge of reach and frequency with sales response. Who can deny that we’re in a mode of continual optimization, where we need the latest results on hand, and update our assumptions with MMM as results become available. Without dynamic access to the MMM model data over the life of campaigns advertisers can be left with a static plan that is unresponsive to shifts in the economic environment, the competitive market, or even senior stakeholders’ priorities.

For all these aspects of the MMM process, different vendors will have different approaches, so it's important to ask questions. All models have strengths and weaknesses. It is important that your partners in MMM and other marketing measurement are fully versed in all the intricacies of applying complicated analytics so that you achieve the ultimate desired outcome:  Spend just what you need to, where and when you need to spend it, to achieve optimal marketing performance.

In summary, MMM models are back, but they're dealing with a much more complex world than before. They can be essential tools in our marketing measurement toolkit. We need to acknowledge their limitations, set boundaries, use new technologies dynamically to simulate, optimize and grow companies’ bottom lines as a result.  The right partner can leverage the power of superior analytics, to realize the full predictive power of MMM in a complex modern media world. 

BIO:

Mark Garratt

Partner and Co-founder, in4mation insights

Mark Garratt is Partner and Co-Founder of in4mation insights (i4i), a marketing performance measurement and optimization consultancy. To learn more, join the upcoming MediaPost webinar co-hosted by Mark and John Nicholson (in4mation insights) with featured guest, Tina Moffett Principal Analyst at Forrester. Register here