Marketing Modeling Mix and Linear Regression

Today’s marketing industry is highly fragmented. Companies no longer communicate to consumers through just one communication medium, for example television or newspaper advertisements. Marketers are now challenged with employing intensive, integrated marketing campaigns that involve media buying, search engine marketing, promoted tweets, blogger outreach, and so forth. The list can become quite exhaustive.

Each campaign component must work together to effectively promote the brand and product, while making the best use of the marketing budget. When employing these campaigns marketers are often challenged with the question of how to effectively divide the budget between each of these communication mediums. For example, would spending additional money on social media versus television advertisements increase sales?

With the business analysis tools available today companies can now create models that will show the effect each communication medium will have on sales. This modeling can be referred to as Marketing Mix Modeling (MMM). When using big data, companies can use MMM in conjunction to explain what marketing tactics generate the most sales and factors that cause the ups and downs of sale. Some examples of factors that can cause sales ups and down include seasonality, coupons and new campaign launches. In its simplest form, MMM is linear regression.

Through linear regression we can model the relationship between two or more variables such as price and sale.  Linear regression looks at the effect of independent variables, such as price, on the dependent variable, typically sales.

For example, The Walt Disney Company & SAS Institute utilized linear regression to find ways to increase their sales. In Disney’s initial findings they learned that a decrease in price increased sales more than increase in television advertisements. In addition, television advertisements increased sales more than online advertisements. This was all found by creating a linear regression.

Instead of creating a scenario of trial and error to determine optimal spending of Disney’s marketing budget, Disney was able to run a regression analysis to find the optimal spending for each medium of communication to maximize sales.

Derived from the basic linear line formula, y = m*x + b, linear regression utilizes the same basic principle where y is the dependent variable, m is the coefficient (or slope), x is the independent variable, and b is the y-intercept.

In a simplified from of linear regression, Disney’s prediction of sales, based on price, could be used by the following equation: Sales = b0– b1 *Price. This equation depicts for every one-unit increase in price, there will be an equal decrease in sales.

In a more complex and accurate form, the equation could consist of many independent variables. Increasing the number of significant independent variables also increases the accuracy of the regression. A more complex fictional illustration of Disney’s linear regression equation could consist of the following: Sales = 3.279 -0.2*Price + 0.06*TV + 0.002*Online + 0.0016*Print + 0.0007*Direct Mail.

Although the primary use of linear regression in marketing is forecasting sales in response to marketing tactics. Linear regression is highly adaptive and can be used for a variety of facets. For instance, eHarmony uses historical data to create a linear regression to match compatible users based on personality traits.

Since eHarmony utilizes traits people may not even know they have, you might be matched with someone who you never imagined dating. This allows eHarmony to match users based on the score created by the linear regression. As a result of using linear regression, eHarmony has approximately ninety users married a day, equating to over 30,000 eHarmony marriages a year, as a result of utilizing linear regression.

Disney and dating sites are not the only one using linear regression. Harrah’s Casino and Resort uses customer age, gender, average income and where he or she lives to create a linear regression. To do this, Harrah’s predicts how much a gambler can lose and still be happy enough to come back again. This number is called the gambler’s pain point.

In this situation the pain point serves as the dependent variable, instead of sales, while age, gender, income and location serve as the independent variables. If a gambler reaches their pain point, $500 for example, then Harrah’s sends out a luck ambassador to improve the gambler’s experience. The luck ambassador could provide the gambler a free meal ticket or a free night’s stay at the resort, any incentive that will increase the gamblers overall experience and keep them coming back.

From determining which customer is more likely to purchase to the correlation of different products, linear regression is more than the basic linear formula we learned in geometry and algebra.

MMM allows companies to analyze their marketing mix while, linear regression is highly adaptable in any industry, not just marketing. As shown in the examples above, linear regression and MMM can be implemented into any field and can generate substantial cost savings. ​


With contributions by Michelle Hill


Brad is an expert in private label and challenger brand food marketing for Barkley, an integrated marketing and advertising agency. He specializes in working with manufacturing-driven food companies that have aspirations to develop a consumer insight-led strategies to help drive innovation and growth.

Comments are closed.