Five Questions for: Damon Ragusa, CEO of Thinkvine
Q. How has the CMO's job of determining the right marketing mix evolved over the past few decades?
A. The emphasis on mix has heightened over the past 20 years for all the reasons we talk about today: media proliferation and fragmentation, rapidly changing consumer habits, speed of information and communication, and an increased emphasis on justifying marketing budgets through ROI analysis. That means the CMO is much more sensitive to marketing mix … or better be.
Q. Has the increased ability to measure just about everything helped or hindered marketing mix modeling?
A. I believe it's done both. More information is never a bad thing in and of itself, but as the amount of information available to the marketing function has grown exponentially, the technology to distill that information has not. The distance between marketers and their consumers has increased as a result, and the reliance on aggregate models to predict behavior and inform marketing mix has widened that gap. There have been only slight incremental improvements in the way marketing mix modeling and analysis has been done. Econometric models that continue to use a regression-based approach that assumes the past equals the future are still the hallmark of most mix models.
Q. What steps should CMOs take to engage in next generation marketing mix modeling?
A. Begin to ask how the consumer fits into the way marketing modeling is done. Insist that “agility” become part of the playbook within their organization, meaning build into the planning process the notion of revisiting your marketing plan at least quarterly. Having good models is only one step – having efficient and scalable models that stay current is equally important.
Q. Do marketing mixes perform differently by segment?
A. Yes, they do. All media has a local effect and most media today is, at some level, targeted. That is to say there are few mass market channels available. So clearly shifting spend levels around different media will activate different segments of the market differently. We capture these differences because we test marketing plans against a derived consumer set. They are not actual consumers, but are generated to be highly representative of real consumers.
Q. How can a model be representative of a real consumer?
A. When you run a simulation, each agent represents one consumer; those agents behave the way we do. When they wake up, do they go to their job? If they do, do they drive? And if they drive, did they see the billboard? They are mathematically driven avatars.