The ABCs of Marketing Mix Modeling

By Craig Ishill, VP Analytic Consulting

Anyone who has worked in marketing has heard John Wannamaker’s adage, “Half the money I spend on advertising is wasted; the trouble is I don’t know which half!”

However, with Marketing Mix Modeling (MMM), marketers can gain a clearer understanding of how much of a return on their investment is driving revenue, profits or other key performance indicators (KPIs).

What is Marketing Mix Modeling

Marketing Mix Modeling is an econometric analytic approach that typically uses linear regression combined with industry specific statistical techniques to establish a cause-and-effect relationship between changes in marketing, operations, price & promotion, products, macroeconomic and external factors in order to predict the impact of marketing and other commercial variables and plans on sales or other KPIs (i.e. store foot traffic, website visits, etc.).

This technique, combined with rigorous data management and harmonization, leading planning capabilities, simulation and optimization tools, and hands-on consulting, helps executives and teams gain more accurate insights into the efficiency, effectiveness and ROI of their investments enabling them to drive higher sales and profits from their spend. A more limited version of this technique – in which the focus is primarily on media channels, is sometimes referred to as Media Mix Modeling.

Today’s leading marketing measurement solutions provide a more holistic approach to supporting a range of marketing and commercial investment benefits. In the context of this, MMM is a component of a broader commercial measurement ecosystem. This typically includes a Machine Learning-based approach that supports an Agile Attribution methodology enabling rapid updates, including platform, creative, audience and placement-level detail as well as incrementality testing. Agile Attribution is a capability developed by Ipsos MMA that embraces an integrated and unified MMM/MTA set of capabilities supporting not only ROI and MROI metrics, but also “true incrementality” in terms of measuring actual short and long-term sales.

Agile Attribution provides media planners and agency partners with timely and granular reads into marketing activities supporting rapid measurement, evaluation and recalibration. Advertisers seeking to leverage an MMM solution should incorporate capabilities like Agile Attribution in order to achieve key short and long-term benefits that lead to stronger brands and increases in sales.

The Benefits of Marketing Mix Modeling

Marketing Mix Modeling offers brands a variety of benefits, but in general terms, it enables the data-driven, fact-based, predictive planning, measurement, optimization, evaluation and recalibration of marketing and other commercial investments in a timely manner with a focus on “true incrementality.”

Some of the more specific outcomes brands can expect to achieve through a successful MMM and measurement program include:

• Optimized Marketing Spend: Besides determining the optimal allocation of budget across activities, a marketing mix modeling solution can produce cost savings by pinpointing the most effective combination of marketing channels, time periods, markets, campaigns and targets, then reducing investment across less effective executions and prioritizing budget to more impactful ones.

• Maximize Revenue and Operating Profits: Effective market mix modeling programs help identify ‘incrementality’ in terms of programs driving actual, incremental sales, not just ROI and MROI (marketing return on investment). Measuring sales incrementality is important because it helps quantify the true value of investments and helps validate the impact the MMM program can have on the business.

• Advertising Effectiveness: Market mix modeling helps quantify the impact of marketing activities, especially more traditional tactics that are not measurable via traditional deterministic methods.
• Synergistic Levers & Halo Effects: Marketing mix modeling helps identify potential synergies and optimization opportunities across sales and marketing as well as marketing and operational factors, etc. as it seeks to understand not only the optimal combination of investments within marketing, but also across other internal and external factors.

• Consumer Behavior: MMM offers insights into how responsive potential consumers are to different tactics and strategies. Modeled outputs can help identify more responsive customer segments and thus improve audience targeting and campaign optimization.
• Collaboration and Accountability: MMM provides not just leadership, but valuable shared learnings across different teams, which can foster collaboration and a growth mindset across various teams such as finance, marketing and sales. By creating a link between marketing activities and outcomes, MMM creates an objective way to justify marketing spend and strategies to stakeholders.
• Competitive Advantage: Brands that engage with a marketing mix modeling solution can gain a competitive advantage by more effectively reaching their target audience and responding to market changes. MMM also allows companies to benchmark their performance against industry norms and identifies target areas to improve their performance.

• Understand the Impact of Advertising on Brand and the Impact of Brand on Sales: In addition to measuring short- and medium-term impact on sales, MMM can measure the long tail of marketing and its effect on sales and brand equity.

• Speed: Today’s MMM insights that integrate Agile Attribution data can be delivered quickly, often on a daily, weekly and monthly basis, enabling users to leverage the insights at the speed-of-their-business.

Beyond providing guidance on efficiency, effectiveness and supporting revenue growth, MMM can do so much more. Often brands will identify additional use cases and new business questions to explore after reviewing resulting model outputs, such as:

• Tracking quarterly/monthly/weekly commercial business drivers
• Optimizing marketing spend and flighting
• Optimizing product mix and quantifying the impact of new product launches and assortment changes
• Understanding the impact of Operations factors and combined synergistic marketing effects
• Developing optimal pricing and promotional strategies
• Forecasting performance
• Providing input into Demand Planning to guide Supply Chain Planning
• Strategic planning
• Recommending next best actions

As MMM’s ability to measure true incrementality on sales continues to improve as a result of larger, more diverse sets of granular data combined with the analytic capabilities to project and predict with it, leading organizations are adopting it to support better decision-making beyond marketing.

• Marketing/Agencies: Table stakes partnership supporting better campaign planning, execution, measurement and recalibrations
• Marketing/Finance: Collaborative and fact-based scorecards, KPIs and tracking tools leveraging more accurate insights by measuring specific, incremental sales results from previous executions help create a trusted platform for working together to invest more confidently in the brands
• Marketing/Finance/Supply Chain Elements: Improved measurement of incremental sales supports finance connecting the outputs into the supply chain in order to generate better forecasts supporting logistics, manufacturing, distribution, etc.

Implementing Marketing Mix Models

Wide buy-in across varying levels of stakeholders within the organization is critical to successfully launching an MMM measurement program.

• Dependent variable and data collection: To begin an MMM program, you must determine what the model will predict and work backward. This is called the dependent variable, or simply the variable the model will estimate. In a sales model, you then need to collect historical sales data. In a brick-and-mortar store traffic model, you will need to collect historical in-store traffic, marketing and related in-store data. This data is usually managed by your internal data and technology teams – along with other, so called, “first-party” data.
• Independent variable selection: Once you’ve determined your dependent variable, the next step is to identify what are called the independent variables. These are the metrics that we want to use the model to identify the impact of. For example, for retail brands, you can bucket independent variables across groups:  marketing, merchandising, pricing, discounting, operational factors, competitive, and external (macroeconomic indicators, trends, weather, etc.).
• Building and validating statistical models: Once the foundational elements of the model are in place, statistical techniques are used to estimate the relationship between the independent and dependent variables. Linear Regression, Bayesian Regression and Time Series Analysis are a few of the more common statistical methods used in marketing mix models to estimate the contribution volumes that each variable has on the dependent (i.e., Sales).
• Attributing sales contributions: The final component is assigning credit or attributing the incremental sales contributions from all the marketing activities, touchpoints and external factors.
• Optimization: Once the MMM model is built, you can use it to simulate different marketing scenarios and predict their outcomes. You can also use the model to optimize the right budget allocations to maximize marketing ROI. Finally, teams can use the model to identify synergies across various marketing channels for integrated media planning.

Understanding How Data Accuracy Impacts MMM

One of the most underappreciated and overlooked factors in Marketing Mix Modeling is data accuracy. Data accuracy is critical to enabling a successful MMM program. Accurate data leads to reliable models, true measurement of incrementality, precise marketing ROI calculations, correct attribution, informed strategic planning, reliable scenario analyses, and continuous improvement.

When it comes to data accuracy, buy-in from stakeholders and across the organization is essential. Everyone needs to work toward the same goal with each data owner understanding that their domain is a small piece to a larger puzzle that will result in higher sales and profits for the brand.

Some of the ways data accuracy can impact market mix modeling success:

• Model Accuracy: The validity of the model is dependent on the accuracy of the input data. Errors in data can distort the relationships between variables resulting in omitted variable bias – where data can over or under-credit the value of the investment.
• Insight Generation: Accurate data provides a foundation for informed decision-making and actionable insights. It allows key decision-makers to make data-driven strategic choices based on reliable evidence versus assumptions or flawed data.
• Decision Support: Inaccurate data can lead to the adoption of ineffective marketing strategies, resulting in wasted resources and missed opportunities.

• Incrementality: Accurate data is core to being able to properly credit incremental sales to an investment.
• Performance Evaluation: Accurate data is essential for precisely calculating the return on investment (ROI) of different marketing activities. Inaccurate data can lead to incorrect marketing ROI estimates, affecting budget allocation and strategy.
• Optimization: With accurate data, teams can optimize their marketing budgets more effectively by identifying the channels and activities that yield higher returns or bigger profits.
• Forecasting: Accurate historical data improves the MMM model’s ability to make reliable future predictions, helping businesses plan and forecast more effectively.

Navigating an Evolving Industry Landscape

The past five years have seen material changes in the marketing and marketing measurement industry (GDPR, CCPA, ATT, IYKYK). Successful marketing mix modeling solutions have had to adapt to these conditions as well as rapidly evolve client needs specific to measuring their commercial investments. This scenario has created an MMM renaissance of sorts. Advertisers are increasingly replacing multi touch attribution (MTA)  with holistic MMM solutions by incorporating Agile Attribution that align to their business needs and can be integrated across their organizations to create measurable, trusted incremental value.

Lately, even several major ad-tech platforms, who long favored individual-level measurement (often within their own walled-gardens), have recently developed and released their own code libraries for implementing MMM.

The current marketing investment landscape that continues to highlight ever increasing and more diverse marketing investment only reinforces the importance and value that Marketing Mix Modeling can provide. Today’s leading MMM providers  offer  a solution for providing a  more holistic, privacy-safe, as well as a short and long-term view of marketing measurement and effectiveness. The increased speed, granularity and accuracy make them particularly valuable for businesses with complex, omni-channel marketing strategies, as well as those operating in environments where user-level signals are becoming increasingly unreliable. MMM provides insights that extend beyond short-term conversions, helping businesses optimize their commercial investments leading to ultimately making more informed longer-term strategic decisions.