Please note: This content is originally written in German. An automatic translation is shown below.
Note: This content is only available in English. The original content is shown.

Marketing Mix Modelling: What Marketing Spend Optimisation and Lego Have in Common

Marketing Mix Modelling (MMM) is a proven method for optimising marketing spend that has been around for decades. The term was established in 1979, but refers to a wide range of models of varying complexity. Joël Bühler, Head of Marketing Intelligence & Technology in the central marketing team at TX Markets gives us an insight into Marketing Mix Modelling, why Lego helps to understand it and how it is applied at TX Markets .

MMM is often an application of time series regression models, usually with two specific characteristics: First, one tries to take into account a time-lag effect: This is because marketing expenditures are expected to influence consumers' decisions at a future point in time. Secondly, marketing expenditures or contacts, which are an important input for the model, are also transformed to account for the effect of diminishing marginal utility.

The Lego Principle

A Structural Time Series Model is used as the basis for MMM at TX Markets. Many of these time series share common characteristics, such as a general upward or downward trend, repetitive patterns or sudden increases or decreases. Structural approaches to time series explicitly take these features into account by representing an observed time series as a combination of components.

With these models, the breakdown into individual, interpretable components (such as trend and seasonality) is therefore central. The analogy to Lego building blocks is therefore obvious. But what exactly are these components? 

The 4 Building Blocks of MMM
  1. Trend - This component is closely related to brand equity, or the factors that contribute to a continuous strengthening or weakening of the company's success.
  2. Seasonality - These are influences that depend on the weekly or annual calendar. An example of this is the well-known "summer slump", or the different user behaviour during the week vs. the weekend.
  3. External and Internal Factors - Unlike seasonality, this component is less predictable and not periodic. It includes, for example, whether the weather has been good or bad, the competition has launched a new campaign or significant product changes have been made.
  4. Marketing Spend - Arguably the most interesting component. This is more specifically about the effect of marketing spend, as this is quantified in the metric of the defined KPI. 

However, which factors should ultimately be part of the model depends on the target. However, one thing to keep in mind is that strongly delayed effects are difficult to quantify. Example: If a user decides to use a product today because of an ad, this user may generate constant value over the next 2 years. Mapping this value in MMM and attributing it to marketing activities is extremely difficult.

MMM at TX Markets

At TX Markets , a Bayesian Structural Time Series Model is used. The time series components are extremely close to Facebook's Prophet model. More can be read about this in The American Statistician (72,1), Forecasting at Scale by S.J. Taylor and B. Lethman. However, TX Markets uses the Python package PYMC3 for modelling. The modelling of time-lagged marketing effects and saturation make up the secret ingredient. 

So far, the use of Marketing Mix Modelling has shown to improve the performance of digital channels. In this article, we were able to give a small insight into the topic and how it is applied at TX Markets. Many thanks to Joël Bühler for the insight. See below for a list of further reading on the topic.

Read More About the Subject
  • Broadbent, S. (1979). One way TV advertisements work. Journal of the Market Research Society, 21(3), 139-166.
  • Liu, Y., Laguna, J., Wright, M., & He, H. (2014). Media mix modeling-A Monte Carlo simulation study. Journal of Marketing Analytics, 2(3), 173-186.
  • Taylor, S. J., & Letham, B. (2018). Forecasting at Scale. The American Statistician, 72(1), 37-45.
  • Jin, Y., Wang, Y., Sun, Y., Chan, D., & Koehler, J. (2017). Bayesian methods for media mix modeling with carryover and shape effects. https://research.google/pubs/pub46001.pdf

More blog articles

Jobs with TX Markets

We're looking for talents. Always.

Techies, geeks, masterminds or project heroes: at TX Markets, people with passion and skills will find exciting jobs in an ambitious environment.