Multiplicative marketing mix modeling

Marketing mix modeling uses statistical analyses to optimize marketing tactics on sales. Marketing mix has been in the industry for nearly two decades and is still used widely to quantify the sales impact of various marketing activities. The business application of market mix modeling in the digital space is called “attribution modeling.”

Market mix modeling uses historic information, such as sales data and manufacturers’ internal data and uses the analytical technique of regression (in particular linear regression). The prime reason behind opting for linear regression is easy calculations of standard deliverables such as ROI, contribution and year- on-year due-tos (change driven by a variable in the model from one time period to the next).

These model learnings are adopted to adjust and optimize the marketing plan. In some cases, it is also used to forecast sales while simulating various marketing scenarios. The two key areas in which these models have limitations is their focus on short-term sales, thus undervaluing long-term brand equity measures of various media vehicles, and bias in favor of time specific media (i.e., TV advertisements) vs. less time-specific media (i.e., monthly magazine advertisements).

Recently, a multiplicative log-linear statistical approach introduced by Fractal Analytics is being used to represent the business model more accurately. Its accuracy is due to the fact that this approach incorporates the non-linear relationship between marketing activities and sales.

Along with increased accuracy, the log-linear analytical approach can help transform results in a way to make it similar to linear model results for the business. Fractal released a white paper that addresses this need by introducing the new approach along with the relevant calculations for the standard deliverables such as elasticity, contribution and year-on-year due-tos to make it more palatable to a business person.

The beauty of market mix modeling is that it can represent reality in an objective, quantitative way. Multiplicative MMM in its previous form, despite having the necessary framework, falls just short of that goal. The white paper aims to simplify the business interpretations of the statistical analysis for marketing teams in business and analytics. After going through this paper, an analyst will be able to understand, implement and interpret the log-linear modeling results and will be better positioned to consult a business person on the changes/adjustments required in the budget allocated to different marketing vehicles. This will, in turn, help the business obtain optimal returns based on their marketing budgets.

To download the paper, “Simplify business insights with multiplicative marketing mix modeling,” visit: ORMS