As advertising channels continue to expand at an alarming rate, there’s growing demand for centralized, panoptic performance measurement and optimization. And while new technology continues to provide more advanced tools and data sets, one of the core methodologies for understanding multi-channel measurement originated in the mid-20th century.
Growing in popularity in the ’60s and ’70s with the rise of TV and the need to measure its impact relative to channels like radio and print, Marketing Mix Modeling (MMM) became a popular topic. And it’s making a comeback, as marketers seek to understand how to drive the best performance for their campaigns.
Marketing Mix Modeling is an analytical approach that uses large sets of historical performance data, to help marketers understand the relative effectiveness of different channels, digital or offline, and forecast performance. It accounts for external factors such as economics, seasonality, market trends, and campaign-specific elements.
While this may sound like a solution to all future forecasting, there are a handful of things to remember when considering and evaluating Marketing Mix Models.
How does Marketing Mix Modeling work?
Through statistical models, Marketing Mix Modeling looks back on aggregate historical data, such as spend and sales, and external factors, like consumer behavior and competitor activity, to understand where to allocate budgets.
Since this requires historical data, most models require a minimum of a year’s worth, and strongly recommend two or three full years. This data can be daily, weekly, or monthly, however the less granular the data, the more is needed for Marketing Mix Models to forecast.
Large advertising companies like Nielsen, Kantar, and Rocker Box often require more historical data than smaller, more bespoke shops, but their models are also typically more robust and complex. Regardless of the model, the more granular the data the better.
Across almost all Marketing Mix Models, the setup is fairly straightforward. Marketing Mix Models can connect to digital platforms and ingest custom data via CSVs, to help you analyze against almost all types of KPIs. You choose the external factors to add, with some Marketing Mix Models already synced with economic or weather data to aid in analysis, and others requiring you to provide your own.
Once all of this input is added, the model runs the analysis and you’ll be able to start understanding the performance of your marketing mix through a dashboard that has all of your data dimensionalized.
Considerations
Marketing Mix Modeling offers in-depth solutions to many forecasting questions, but may not be the right solution for all campaigns. Aside from the obvious issue of requiring a lot of historical data to get started, these models are not necessarily meant to provide agile solutions to continuously optimize performance.
Refresh frequency
Larger companies often have access to more complex models and other data sets to build out granular forecasts, but then only refresh the model monthly, quarterly, or annually. This means these models not only typically require more historical data, but also that you can’t see the impact of adjustments until much later.
One of the advantages of smaller Marketing Mix Modeling companies like Sellforte, Brunner, and Lifesight is that they offer more frequent refreshes, so you can gain insight and optimize more often.
These models are by no means less effective, and depending on your channel mix, budgets and KPIs, you may find a more agile model with more frequent refreshes can help you implement changes faster and drive performance more effectively than their more complex counterparts.
While these models are more agile, they’re still meant for understanding relative channel performance and forecasting at a macro level, not tracking channel touch points and assigning conversion credit for granular, micro-level optimizations. For that, we recommend Multi-Touch Attribution (MTA) which focuses on tracking channel touchpoints to more accurately assign actions or conversions.
Cost and ROI
In addition to the amount of historical data and refresh frequency, cost is also a major consideration. There is a base cost for the initial build of the model—often based on a percentage of media spend, or a fixed cost for some smaller Marketing Mix Modeling companies, with additional costs for each model refresh.
Expensive models generally offer more robust dashboards, a more complex analysis and more support. But for the most part, regardless of company size, these models are expensive. For a strong ROI, you’ll likely need to be spending a significant amount on the campaigns you’re forecasting.
Conclusion
The flexibility of Marketing Mix Models allows you to forecast almost endless possibilities across most KPIs, but requires a lot of data and a lot of budget. If you’re working with an established campaign and want to gain a deeper understanding of how to plan for the future, Marketing Mix Models are something we recommend you explore.
On the other hand, if you’re ramping up a newer campaign, focused on agile optimization, or don’t have large budgets for testing and analysis, there are other considerations that may be more impactful to your campaigns than Marketing Mix Modeling.
FWD has helped launch, optimize, and scale numerous omni-channel campaigns across healthcare, animal health and non-profit. We can help you move your channel analysis forward, no matter the datasets or the budget you’re working with.
Check out our work and get in touch with us.
