Make Smarter Marketing Decisions with Media Mix Modeling

Every brand wants to pre-optimize their campaigns for success right out of the gate, with parameters in place to quickly attribute success to specific channels, audiences, and creatives. However, as we look toward a future without the simple attribution offered by cookies, we need to get back to our statistical roots of traditional media strategies, those without the easy button involved. That’s why marketers should lean further into the value that’s offered with media mix modeling.

What Is Media Mix Modeling?

At a high level, media mix modeling is a way to define optimal budget allocation for media channels by looking at previous campaign performance. It requires analyzing sales-related data and media metrics (Coegi’s data and technology team recommends 2-3 years worth) to make predictions and strategy optimizations that will, in theory, improve future campaign performance. 

Is Media Mix Modeling Superior to Attribution Models?

In short, these models should be considered and analyzed alongside each other, as they both offer valuable insights from unique perspectives.

• Attribution gives you quick, real-time information about how specific media parameters are impacting your business goals. This information is useful when making mid-flight optimizations and short-term reporting.
• Media mix modeling zooms out to give a bird’s-eye view of how all the pieces are working together to affect long-term strategy and performance. Each model informs the other, but tells different stories. 

How Can Marketers Build Strategies from Modeling Learnings?

It is crucial to fully understand the data you’re analyzing, not just the standard media metrics from campaign reports. What are all of the factors that may have contributed to performance fluctuations?

• Creative?
• Messaging?
• Audience strategy?
• Seasonality?

Knowing the context surrounding the numbers will give you a strong foundation to build future strategies upon. Using that context as the framework, determine what story the data points are telling you. The numbers don’t lie, but they don’t always tell the whole story. By asking the right questions, and maintaining a test and learn mentality, you will ensure strategic decisions are based on multiple factors rather than just one KPI.

How Do You Know Your Media Mix Model Is Working?

A media mix model makes predictions, but it’s not a crystal ball. Just because Facebook historically performed well does not mean it will continue to do so forever. That said, it is important to develop a continuous learning agenda to design your models.

Test your assumptions based on historical performance. For example, what will happen if you increase the budget for programmatic channels? Do overall business results change? The only way to know is to strategically make the budget adjustment and measure incremental results. From there, you can make more informed decisions about your channel strategy and budget allocation. 

Priming Your Media Mix Model For Success:

• Keep business goals at the center of your strategy 
• Gather quality, historical data to measure actionable results 
• Understand the contextual factors impacting your data results 
• Consider the sales cycle when designing tests – a longer cycle needs more time between strategic adjustments
• Share strategic details and learnings across teams. Seemingly trivial aspects to you may impact how a model is built. 
• Data can be easily manipulated to tell an inaccurate story. Think critically and apply business acumen to make sure you have sound methodology. 

Further Reading:

• How Curiosity can Improve Your Data Analytics
• How Many Paperclips Can You Fit in the Internet?
• Measure What Matters