Marketing Mix Modeling: a powerful tool that is becoming increasingly relevant
Guilhem Bodin, partner and media expert at Converteo, shares his insights on Marketing Mix Modeling (MMM), a powerful tool for advertisers that is becoming increasingly relevant in the current context of diminishing third-party data and the questioning of deterministic performance measurements.
—
Key Takeaways:
- Marketing Mix Modeling (MMM) is becoming increasingly relevant for brands due to two converging trends: the scarcity of personal data following the implementation of GDPR, which is offset by easier access to operational data through the “dataification” of marketing.
- When properly deployed, MMM is a reliable and powerful decision-making tool that transforms marketing into a profit center rather than a cost center. It also offers numerous advantages in the current context, including being unbiased (unlike media tracking solutions) and respecting consumer privacy.
- Technologies are rapidly evolving to provide greater flexibility and agility in using models. Platformization, for example, enables near real-time data injection and analysis. This industrialization of MMM projects is a valuable lever for increasing responsiveness and performance.
—
For several years, the marketing and media buying world lived with a fantasy: the idea of a world where everything, or almost everything, would be “trackable” and measurable. While this is entirely possible in some countries, GDPR in Europe has effectively buried this vision. In this context, statistical modeling—particularly Marketing Mix Modeling (MMM)—provides an interesting alternative for the marketing departments of major advertisers.
Specifically, through mathematics, Marketing Mix Modeling measures the impact of marketing factors on performance. In other words, its goal is to evaluate the role of marketing in relation to a KPI to be explained.
The model can include explanatory variables from the marketing mix (media, pricing, promotion) as well as external factors (weather, events, seasonality, market-specific events) and other possible factors, such as logistical issues. As long as there is sufficient historical data on a factor, it can be integrated into the model.
However, it is important to note: an MMM model analyzes the impact of a selection of explanatory variables on ONE dependent variable. To study multiple variables, multiple models should be combined. For example, a model aiming to explain revenue will yield different results from a model aimed at explaining changes in a brand awareness indicator. Analyzing the results of these models allows for decisions to be made with a comprehensive view of the role of marketing factors.
A powerful tool for understanding and optimizing
When well deployed, this tool becomes a powerful lever for understanding the impact of pricing, promotion, or a loyalty program, as well as a communication channel, an advertising message, a type of targeting, or a media buying strategy. It thus allows for estimating the impact of changes in endogenous factors, such as marketing mix, media mix, or media tactics, to adapt future marketing strategies accordingly.
For example, in the case of a retailer using numerous marketing channels (press, TV advertising, digital advertising, loyalty, promotion, etc.), MMM allows for evaluating the impact of each investment on revenue, incremental sales, store traffic, or brand awareness, depending on the chosen KPI, with the goal of optimizing marketing efforts to improve performance.
Numerous Advantages in the Current Context
In summary, Marketing Mix Modeling (MMM) is a highly relevant solution for evaluating marketing impact, at both strategic and tactical levels, whether for managing investments in detail or making large budgetary decisions. By enabling the rationalization and measurement of marketing efforts, it paves the way for a significant shift for marketing departments: transforming them into profit centers rather than cost centers.
This method offers several advantages, with its unbiased nature being a key benefit: MMM does not rely on attribution models or biased tracking methods. It also adheres to “privacy-by-design” principles, as it allows for the analysis of activation roles without depending on legal and technological contexts (ITP/ETP/Adblockers/GDPR) and individual tracking of interactions.
While regulatory and technological constraints limit access to personal data, marketing has, conversely, become increasingly “datafied” in recent years. Access to data is now simpler and more automated, and the creation of centralized marketing data warehouses allows for more automation in the processing and utilization of marketing data.
Platformization to Facilitate Data Injection and Analysis
This trend is significantly enhanced by the “platformization” of MMM tools – similar to the MMM platform we developed at Converteo – which allows for the implementation of detailed, regularly updated models. The most advanced companies have already industrialized MMM and are now capable of evolving their models at increasingly frequent intervals (quarterly, monthly, or even weekly), by injecting their data in near real-time! As a result, the models are becoming more agile and adaptable, leading to a practical application of MMM in decision-making.
However, for an MMM tool to realize its full potential, marketing media and acquisition teams must be able to align strategic and operational visions, meaning they need to translate the model’s results into clear recommendations that can be directly applied on the ground. As often the case, the tool alone is not sufficient; the accompanying support and operational advice are crucial.
Converteo’s MMM Factory Platform
Curious about how our Marketing Mix Modeling expertise can elevate your marketing and media campaigns? With Converteo’s MMM Factory, easily model the incremental performance generated by your marketing and media actions on your economic performance. Managing your marketing and media budget, reducing costs, and maximizing the performance of your investments has never been easier.
