MARKET MIX MODELING: AN INTRODUCTION

Market Mix Models (MMMs) are not a new analytics practice—they have been around for about 30+ years. However, MMMs—which lost some of their sexiness along the way—are back in vogue, having continually been the gold standard of measurement, largely due to signal loss and the upcoming end of cookies.

An MMM is a statistical analysis using a multi-variate regression model against a dependent variable or KPI (such as sales). What does that mean? Think of a cake.

An expert baker can taste a cake, created by someone else, and determine the ingredients and their proportions: approximately 30% flour, 15% sugar, 5% vanilla extract, and so forth. An MMM is that baker—breaking the “cake” (typically sales) into its components, while understanding the proper value and mix of those components with a high degree of confidence.

An MMM certainly measures the impact of media—what percentage of sales are attributed to YouTube efforts, as an example—but also goes far beyond that, to understand the impact of competition, COVID, price changes, product launches, gas prices, seasonality, and more. For a time, MMMs were overshadowed by real time data with digital touchpoints—KPIs measured by Meta or Google Analytics; MMMs are not extremely real-time. Signal loss—largely due to iOS 14.5—has pushed companies to again rely on MMMs as the way to determine what is driving their business. As certain KPIs are increasingly difficult to get from digital channels, MMMs offer a statistically robust, full funnel view of performance, a picture broader than digital channels could ever deliver.

REALLY UNDERSTAND THE IMPACT

MMMs go further than breaking down sales by drivers, they also measure decay and diminishing returns for media. We know media can have a carryover impact on purchase behavior—and that the seventh media exposure to a consumer, for example, will not have the same impact as the first. MMMs analyze carryover impact by channel, so a client can understand that TV is having a stronger lingering impact than radio, as another example, guiding that client’s strategy on the go-forward.

Another way that MMMs can influence future planning: at a certain point, effectiveness of media channels can plateau and spark diminishing returns. We use the output of MMMs to fuel our response curves tool, Spark, to help clients optimize their media budgets and predict when certain media channels will have exhausted their investment.

QUESTIONS MMMs CAN ANSWER

• What are my media channel and non-media contributions to the KPI? (A KPI example is sales, and those sales can be ecommerce, in-store, or combination)
• What are my media channel ROIs?
• How can I optimize, and scenario plan my media mix?
• What is a baseline level of sales I can expect if I turn media off?

Keep in mind, MMMs—like all other sales and analysis techniques—cannot provide insights into channels with no history. MMMs also will not dive into micro-performance of media, such as creative performance within channels.

RUNNING AN MMM

An MMM requires a good amount of historical data; depending on the KPI you are trying to measure, you will need several years of historical data on that KPI, ideally broken out by week. We typically run a feasibility analysis to make sure that a brand is a good fit for an MMM—or if another advanced analytics tool might be a better fit. Single channel brands are often better suited to a Match Market Test, as several channels better substantiate an MMM.

As we navigate changes brought on by iOS14, MMMs are an incredible solution for brands to understand, with great detail and confidence, what is impacting their business. Expert analytics helps build excellent, foundationally strong brands. If you are interested in powering your brand with expert data and informed strategy, connect with us.