ARF Forum: Marketers Weigh Impact Of Marketing Mix Modeling

ARFartAt the Advertising Research Foundation’s Industry Leader Forum yesterday in New York, one point of contention was around marketing mix modeling.

Marketing mix modeling is “a core part of our decision-making,” remarked Patrick McGraw, director of consumer and market at P&G, during a panel about measuring marketing effectiveness. “It helps us develop a better understanding of what’s happening, what isn’t and deliver better value to consumers and better shareholder return.”

But, according to McGraw, something’s missing. On the one hand, there has been a surplus of talk about data integration, but few answers to “how we bring analytics together that builds on the backbone of our marketing mix model.”

Another area for improvement, he said, is taking a “total, comprehensive business” look at the marketing mix model. “Much of the focus has been on media and advertising, but [we need to look at] how we drive business growth overall and pull all the levers in our mix.”

At ESPN, the “longitudinal” aspects of data and modeling – “being a historian and determining results year-over-year” — are important, said Julie Perlish, senior director of advertiser insights.

For General Mills, marketing mix modeling provides a “common framework” across marketing, finance and operations, according to John Walthour, director of global consumer insights for General Mills. His team, going by unit-name Business Driver Analytics, is responsible for measuring the direct interactions between paid, earned and owned media at the brand level, as well as the organizational level.

Panelists fielded a question about the biggest challenge for marketing mix modeling, with one audience member citing the CFO and overall accountability if the mix model is unable to prove out or otherwise properly attribute data outcomes.

“Finance,” responded McGraw, “is a critical partner for us. Sometimes, the results seem different from what we might have anticipated.”

As Pat LaPointe, EVP of MarketShare Partners, later pointed out, there is no such thing as “a model” anymore.

Marketers need to look at broader market factors like mergers and acquisitions, short-term revenue vs. long-term brand effects and the customer lifecycle in its entirety. “We’re building a whole bunch of little models that talk to one another,” LaPointe said. “But any time we try to build [data] optimization models [on a] large number of variables, we need the horsepower” that advanced analytics and technology bring.

At one point, the panel digressed into a quasi-debate about the correlation component of marketing mix modeling vs. causality; an audience member called out that critics of the marketing mix model are, inherently, stakeholders that could potentially take a budget hit if attribution/mix models dictate that action, should they claim that statistical regression is flawed.

“I think the key to having success is being closely aligned and integrated with the business teams, which is true not just with marketing mix modeling, but any analytics,” Walthour remarked. Thus, keeping revenue management, finance and operations included in the marketing conversation, could make for more effective execution.

Enjoying this content?

Sign up to be an AdExchanger Member today and get unlimited access to articles like this, plus proprietary data and research, conference discounts, on-demand access to event content, and more!

Join Today!

1 Comment

  1. Sami Assili

    Has anyone used Neural Networks or other non linear/black box techniques in MMM? If yes, then how do you arrive at the weights/coefficients for each of the variables like TV, Print, Radio, Digital etc for computing contributions for sales decomposition or for computing ROIs or for making future predictions for optimal spends using the F(x) for Sales… Am curious about the application of such techniques in MMM, hence any inputs/suggestions would be much appreciates.