What You Should Be Getting From Marketing Mix Modelling (But You’re Probably Not)
By William Marks, Mutinex
In the context of a changing reporting landscape and enhanced modelling capabilities, marketers are seeing more potential in MMM for reporting than ever before.
Marketing mix modelling (MMM) is an old technique that’s been given new life in recent years by two things.
The first is the imminent death of the cookie and the associated decay of multi-touch attribution (MTA) models. For the first time since the introduction of performance media, marketers are struggling to attribute sales clearly. This has a flow-on effect for managing marketing reporting and budget planning cycles.
The second factor is the increased computing power at our fingertips which allows for faster, more intelligent modelling. This unlocks the potential of MMM in new and interesting ways that marketers should be leveraging. So here’s what you should be getting from market mix modelling (but you’re probably not yet).
The need for speed
What’s different about MMM now that we have better access to computer processing power and friendly SaaS interfaces to deliver results? Primarily, the answer is speed. Companies that are getting the most from their MMM have identified a need for a regular stream of data and insights via MMM that it isn’t possible to garner from half-yearly or even quarterly reports.
This is because it’s difficult for marketers to get stakeholder buy-in to make changes to marketing investment patterns off the back of a report that is delivered so infrequently. Many Mutinex customers have reported that models that deliver on a longer cadence are useful for only one function: reporting on past marketing performance.
“What happens in lots of businesses is that these things land as a PDF or PowerPoint and can get lost in someone’s email – and after a while, you easily lose visibility of them,” says Carl Bunn, Head of Data and Solutions at Samsung Australia.
Reporting on past performance is a good use for MMM. We have observed that many of our customers find that reporting on past results using MMM is an effective way to “reset” the conversation around marketing budget from expenditure to investment and drive some confidence in marketing decisions moving forward. But reporting on value already driven is just the first way you can use next-generation marketing mix models.
The growth flywheel: Test, learn, iterate
MMMs that deliver results on a monthly cadence are extremely powerful because they allow marketers to put growth results at the centre of the planning cycle instead of campaign results. And they can do it without making large (and for some companies seemingly sudden and possibly problematic) changes to marketing investment.
We encourage all of our customers to adopt a test-and-learn approach that starts with considering ROI on tactics at a channel level (perhaps the ad format for example). Then we measure the impact of small changes over the course of a month (or a campaign period), before iterating on those changes to compound the effect. We call it the ‘growth flywheel’ and it’s made possible by higher model refresh frequency and more granular insights produced by modern MMM.
For example, Mars took a careful test-and-learn approach to the implementation of MMM in its Australian confectionary business in 2023. Working with the Mutinex team, it was able to first isolate channel ROI. Starting with online video, the team decided to reduce investment in underperforming formats. In the space of the first month, the team doubled ROI on online video and generated a six-figure increase in revenue based on that one experiment alone. Compounding results over time and across different channels has delivered an impressive uplift to marketing ROI across the course of the year.
Marketers that have this process running smoothly find that it scales easily to channel spend, interrogation of channel mix and then, finally to total media budget. The big advantage of working in this way is the credibility marketing buys within the business for taking an ROI-driven view of planning and marketing output.
“What we’ve seen with [monthly modelling] is that you have models that actually empower you to make decisions because they’re dynamic,” says Megan Quinn, Head of Data and Digital Products at Asahi Beverages Australia.
“The more robust the data, the more powerful the platform becomes,” Quinn added. “This is growing confidence in the outputs and supporting a new way of thinking about the true power of marketing.”
Getting more juice from the data lemon
The big difference between past iterations of MMM and what’s available for marketers today really comes down to one thing: the speed at which your model can deliver data and insights. In the context of a changing reporting landscape and enhanced modelling capabilities, marketers are seeing more potential in MMM for reporting than ever before.
But whilst MMM can certainly be used as a reporting tool, if you can only use your model outputs to identify and prosecute value already created, then you’re only squeezing about 30% of the juice from your data lemon. The remaining 70% of value comes from using your model to plan, test, learn and iterate on your marketing investment to compound your growth trajectory. This is what modern MMM unlocks for marketers.