Modern Marketing Mix Model - Next Advertising Era
Let's explore what they are and how they might replace the Multi Touch Attribution (MTA) models, possibly reaching a dead end due to new privacy regulations and the disappearance of 3rd party cookies.
The sunset of 3rd party cookies will limit the ability to track post-impression user interactions, and privacy protection restrictions will limit the ability to accurately track the user's various interactions with the company's digital properties. As we have seen in previous episodes, Google's Attribution API will partly recover this information, but the analysis capability is already limited by the inability to track impressions from social networks like Meta and TikTok. The reality is: the sunset of 3rd party cookies has only given the final push to a sector that needed to rethink itself for years.
The industry has continued to seek solutions, finding in the old Marketing Mix Models a possible answer, but what are they?
What is a Marketing Mix Model?
A MMM, or Marketing Mix Modeling, is an analytical method used to measure the effectiveness of different marketing activities. It uses historical data to understand how various components of the marketing mix — such as advertising, promotions, pricing, and distribution — impact sales or other objectives. Essentially, a MMM assesses the contribution of each marketing channel and tactic to the overall ROI, allowing marketers to optimize spending and strategy for future campaigns. The data used in a MMM can vary widely but generally includes:
Spending Data: Marketing costs broken down by channel (e.g., TV, radio, social media).
Performance Data: Sales or other relevant KPIs, such as leads generated or web traffic.
Media Data: Impressions, reach, frequency, GRP (Gross Rating Points) for advertising campaigns.
Economic Data: Macroeconomic indicators like GDP, interest rates, consumer data.
Competitor Data: Marketing activities or promotions of competitors.
External Data: Events, holidays, weather conditions that might influence consumer behavior.
Digital Data: Click-through rate, conversion rate, website navigation data.
These data are analyzed to understand the effectiveness of different components of the marketing mix and to assess future strategic decisions.
What characterizes a Modern Marketing Mix Model?
Modern Marketing Mix Models (MMM), such as Meta's Robyn and Google's LightweightMMM, represent progress in the field of marketing analytics because they leverage artificial intelligence and machine learning to optimize advertising spending and underlying strategy. Robyn is an open-source package that uses various techniques of multi-objective evolutionary algorithm for hyperparameter optimization, time series decomposition, Ridge regression, and gradient-based optimizers for budget allocation. It is particularly suitable for digital and direct response advertisers with granular datasets and many independent variables.
Google's LightweightMMM, on the other hand, is a lightweight Bayesian MMM library that assists companies in understanding and optimizing marketing spend across media channels. The Bayesian approach integrates prior information into modeling, allowing for the use of industry experience or previous models. It also offers hierarchical modeling options and is designed to accommodate media saturation complexity and delayed effects, capturing the lagged impact of media channels on sales.
Compared to traditional MMMs, which focus on measuring the impact of advertising and promotions, modern models like Robyn and LightweightMMM benefit from automation and advanced data processing capabilities. They can handle more complex datasets and provide more nuanced insights into the effectiveness of digital marketing channels. The use of AI reduces human bias and can democratize modeling knowledge, empowering companies of various sizes.
However, the increased complexity and sophistication of models like Robyn and LightweightMMM can be a disadvantage. They may require more expertise to be correctly interpreted and potentially lead to an over-reliance on machine-generated insights, which might not always account for the unpredictable nuances of consumer behavior or market changes.
On the other hand, traditional MMMs, although they may not have the same level of granularity or automation, offer proven and familiar frameworks to many organizations. Their predictions are often easier to understand and give greater confidence to stakeholders, a factor not to be underestimated in encouraging different departments to adopt the generated outputs.
In summary, modern marketing mix models offer powerful tools for optimizing marketing spend, particularly in the digital and direct response advertising space. However, ensuring that the insights are actionable and building trust in these models is crucial for successfully integrating them into marketing strategy and operations. Traditional MMMs, although less sophisticated, provide a level of reliability and comprehensibility that continues to be valuable for many organizations.
What are the main differences between the traditional approach and the modern approach?
In traditional MMMs, the results typically show the overall incremental impact of marketing investments and predict outcomes based on historical data. For example, a traditional MMM might indicate that for every $100,000 spent on TV advertising, there is an average increase of $1 million in sales.
Modern MMMs like Robyn or LightweightMMM can provide more detailed results thanks to their advanced algorithms and machine learning techniques. They might reveal, for instance, that the incrementality of TV advertising varies weekly, depending on the type of creative used or the presence of concurrent online campaigns, and they might also offer optimization strategies for budget allocation across channels in a more timely manner.
The key difference lies in the level of detail and the ability to adapt quickly to change: modern MMMs offer more dynamic and complex insights compared to the general trends highlighted by traditional models, thanks to the possibility of updating them in very short periods, even weekly.
What are the differences between Multi Touch Attribution and Marketing Mix Model?
The main differences between modern Marketing Mix Modeling (MMM) and multi-touch attribution (MTA) models lie in their goals and the types of data used:
Objectives: MTA focuses on the impact of individual touchpoints on conversions and sales, attributing credit to multiple touchpoints in the user's journey. MMM focuses on the overall impact of a company's marketing mix on its sales and other outcomes.
Types of Data: MTA uses granular, device-level data to track user interactions with various touchpoints. MMM uses aggregated campaign- or channel-level data, making the impact of individual touchpoints on conversions and sales less visible.
Advantages of MMM: It is privacy-oriented, uses aggregated data, and can provide predictions on the impact of marketing strategies and aid in optimization, identifying the most effective marketing channels to enhance.
Advantages of MTA: Offers greater granularity and can be combined with incrementality measurement to make causal inferences, helping to better understand the effectiveness of individual touchpoints and thus optimizing marketing campaigns.
In this table, I summarize the characteristics of the two methodologies, drawing inspiration from a table created in an Airbridge article.
Robyn and LightWegth are in their early stages, but are becoming increasingly promising. There are various vendors of Marketing Mix Models on the market that have even more advanced methodologies and offer SAAS solutions. The list is not exhaustive; these are the producers whose updates I follow. The description was generated by ChatGPT based on the homepage of their websites.
Forvio is a SaaS (Software as a Service) tool for Marketing Mix Modeling (MMM) supported by artificial intelligence. It helps reduce human biases, providing actionable decisions based on data and science. It aids in revealing the real impact of marketing, discovering media saturation points to optimize spending, and learning from past activities for continuous improvement. It is optimized for teams of all sizes, from freelancers to large corporations, and provides in-depth insights for efficient budget management and an effective marketing strategy.
Adtriba is a data-driven marketing platform that offers tools for measuring and optimizing marketing activities. Through automated data collection and AI-based evaluation, Adtriba provides actionable feedback for budget allocation and strategic alignment. The platform eliminates partial analysis and provides a comprehensive understanding of marketing activities, both online and offline, enabling more informed decisions and optimizing ROI. Adtriba Core focuses on multi-touch attribution and unified marketing measurement, while Adtriba Sphere specializes in Marketing Mix Modeling.
Rockerbox is a data-based marketing measurement and analysis platform that allows combining different measurement approaches to best suit specific needs. It offers solutions like marketing mix modeling for budget planning, multi-touch attribution for attributing sales to various marketing touchpoints, and tools for daily campaign optimization. Rockerbox helps unify marketing data, balance privacy standards, predict future performance based on past results, reduce inefficient marketing spending, and bridge gaps in attribution models, providing a complete view of the customer journey.
Cassandra is a machine learning-based Marketing Mix Modeling software that analyzes marketing data to detect budget waste and provide media plans that maximize ROI. The ROI optimization process occurs in three stages: identifying inefficient campaigns, simulating and forecasting the best media plan, and reallocating the budget to maximize impact.
Recast is a Marketing Mix Modeling software that uses Bayesian analysis and automated data pipelines to optimize marketing performance. It measures impact in real-time without tracking bias, providing reliable estimates on ROI and channel saturation. It offers weekly results and tools for optimized budget allocation based on specific goals, helping marketers invest precisely and without waste.