More than 25 years ago, MMA pioneered the use of marketing mix modeling and today remains one of the industry leaders in the space across a wide range of industry verticals. We’ve extended our core marketing mix capability (which is now known by many clients as Commercial Effectiveness) by integrating data sources specific to the major industries we serve, and a wide range of digital and social media captured through DMPs, on-boarders and directly from publishers to reflect the changing marketing and media environments around us, and the consumer behavior which results. I’ve also had the responsibility and privilege of advising senior decision makers—CEOs, CMOs, CFOs–on marketing investment choices for almost as many years across some of the world’s leading companies and brands, and marketing mix modeling has always been a core part of my analytical tool set.
I’ve been a client and a competitor of MMA; now I am proud to be a part of the Company’s leadership. I have had to defend marketing mix as a tool in prior lives because it was simply the right thing to do to help move the business of my respective company forward. Decision makers often faced tough choices moving tens of millions of dollars (hundreds of millions of dollars over time) across marketing strategies, tactics, campaigns and even operations’ factors to move their business forward and meet their responsibilities to their shareholders. And I am defending marketing mix modeling now because it is still the right core capability to support moving the business of my clients forward. But it is also my business—MMA’s business—that I am defending now as well.
The critique of marketing mix modeling posted on LinkedIn on 8/31/16 by Michael Wolfe—The Death of Marketing Mix Modeling as we know it—fails to reflect current industry practice, advancements in data, or modeling approaches, and a general understanding of how this capability has evolved to address the latest business questions and marketing ecosystem. It’s misleading. And it fails, as I have so often seen these so called critiques fail to do, offer practical business planning alternatives.
Overall, an Outdated, Invalid Perspective on Marketing Mix
Mr. Wolfe’s posting cites research that is 10+ years old, and focused on only the CPG industry. So the fundamental validity of Mr. Wolfe’s premise–generalizing to a dramatically changed business context of 2016–is called into question. Consider for example that the 2006 foundational reference (which I believe is really a 2005 HBR article) relied on data before Facebook became available to the public (around 2006), the launch of the first iPhone in 2007, and the launch of YouTube in 2005.
The discussion also hinges on a relatively narrow industry lens of Consumer Packaged Goods which could lead to unnecessary concern that TV and Media are being “short-changed” in ROI measurement across other industry verticals where we have observed and measured very strong returns. All industries and product categories possess unique business model characteristics that underpin profit margins overall, and returns on marketing in particular. The ROI construct, and the range of observed ROIs, can vary substantially across industries.
Measuring Short and Long-Term Effects, When they Matter
Mr. Wolfe also states that ” MMM [marketing mix modeling] focuses on the short-term effects of media and generally ignores or does not measure the long-term effects.” MMA’s core marketing mix modeling methods do in fact focus on the short term drivers but we also enable measurement of long-term marketing and media effects through our Brand Funnel measurement capability which integrates marketing mix and brand equity modeling. The relevant question, however, for an advertiser is what will impact year on year plan changes?
Contrary to Mr. Wolfe’s Article, Marketing Mix Models Do Measure Creative Differences. And message length. And dayparts. And…
Mr. Wolfe holds that “MMM models only measure the impact of ad GRPs or spend and not the ad message or creative.” This is stunningly false, and has been an industry red herring for years. Marketing mix models have long accounted for differences in creative by utilizing separate model variables for different campaigns and creative executions. Specifically, GRP streams are separated by relevant campaign or copy creative, transformed individually, and tested for significance—allowing each material creative execution difference (if supported by adequate in-market weight) to receive its own effectiveness (and ROI) read. At MMA we do this regularly, as well as testing for differences in Dayparts, :30s vs. :15s, difference across DMAs, etc. The choices in what insights to focus on in any model are driven by client key business issues. But the idea that marketing mix models cannot capture differences in executions by copy, across channels, or mediums, is simply false—and has been for a long time. In addition to the TV example above, we go even more granular on paid, owned and earned digital media by reporting results at the sub-channel, campaign/creative, placement/publisher, timing and device level that enable tactical optimization to improve the ROI of each marketing channel.
Ironically, Many Digital Attribution Models Don’t Account for Attribution Bias, and Marketing Mix Modeling often is the Holistic Option
By capturing holistic marketing effects and the “Net Attribution” of each marketing channel through a range of multi-stage models that require that sales contributions actually add to 100% (on and offline), our digital attribution models actually provide a complete, financially sound planning base for investment decisions. So MMA’s models, at least, are not “confused”, when consumers are “affected by a pathway of touchpoints” described by Mr. Wolfe. Our models capture the cross-media effects, identifying and quantifying synergies and interdependencies across all relevant channels. This synergistic understanding of cross-media effects, across a wide range of communication platforms and marketing tactics is a key part of what has evolved marketing mix toward “commercial effectiveness” modeling—discussed further below.
The Voice of the Customer is Easier than Ever to Incorporate into Marketing Mix Models, in More Ways than Ever
Marketing mix models leverage a wide range of data sources, including search and social media outlets, customer satisfaction as well as individual customer level data that enables optimization at the customer segment level and represent the “voice of the customer” called out as a missing element by Mr. Wolfe. Additionally, MMA has broadened its definition of marketing mix to include operational factors that are especially important to restaurants and retailers—such as operating hours, manager and employee tenure, guest satisfaction data, and more—that impact and reflect customer experience; we also include salesforce factors that are critical in pharmaceutical commercial effectiveness modeling. In short, marketing mix modeling at MMA has been broadened to reflect “business” or “commercial” mix modeling that accounts for a total go to market model, which encompasses the widest range of customer experience factors possible.
Alive and Kicking
Marketing mix modeling is alive and kicking because marketers and business leaders continue to need holistic, multivariate answers to fundamental business questions. Businesses need math to add to 100%. The science and the art—despite the periodic under-informed industry dust-ups—continues to progress.
I’m glad to defend this work. It’s been much of my life’s work. It’s my business. It’s MMA.
To learn more about Patrick McGraw, please read his bio here.