Original blog post: https://sellforte.com/blog/why-you-should-use-continuous-marketing-mix-modeling/
MMM has been traditionally implemented as externally assisted projects, which typically include either a media or consulting agency that’s responsible for the modeling.
This approach ensures that the modeler (agency) has enough time, resources, and prioritization from the client company to deliver top-class modeling results. Which at first glance this seems like a win-win situation, right?
Despite the great premise, MMM projects have built a bit of a reputation for themselves: They’re either extremely long and expensive consultancy projects or quick-and-dirty modeling sprints by media agencies. And both cases are suboptimal for the client.
On this basis, Sellforte was founded.
How MMM creates value for the client
MMM has two primary use cases that make it extremely valuable for marketers: (1) Test and (2) build hypotheses about your marketing effectiveness:
MMM validates whether your current campaign and media mix has been as effective as thought to be when planning the campaign
MMM indicates optimization potential by revealing campaigns and channels’ incremental uplifts as well as the diminishing return curves
If we look at how MMM creates value, both consulting and media agency projects have their benefits and disadvantages:
Long and thorough modeling projects provide comprehensive validation for the existing hypotheses, but poor hypothesis test & learn capabilities due to the length and cost of the project
Agile and light modeling projects provide quick test & learn capabilities, but the modeling results are less accurate and might not validate the hypotheses that well
We can quickly see that combining the best of both worlds would be the best way to create a maximal amount of value for the client: Clients would get constant yet credible validations for their marketing effectiveness questions and hypotheses.
To achieve this the modeling needs to be scientifically top-notch, and at the same time it needs to provide new insights as fast as possible.
On a continuous basis.
Why there hasn’t been continuous MMM available until now
It’s always easy to say how things should be.
Even with a simpler setup, MMM requires a lot of data and analytical expertise to connect the activities with the results accurately and realistically.
More advanced MMM methodologies, such as Bayesian Hierarchical modeling which is the gold standard in MMM, require even more as the modeler needs industry expertise in setting up the business priors that guide the models to find the right results.
The complexity of MMM has previously led to a situation where it’s either too costly to run the modeling on a continuous basis, or where the results are too high-level/low-quality for ongoing optimization.
So, what has changed?
In a nutshell, the latest technological development in cloud computing and probabilistic programming tools enables experts to build highly automated and scalable models.
Combine the exponential computing power with data connectors like Supermetrics and Funnel.io that can standardize the data formats and automate the data imports, and you get continuously updated MMM results with uncompromised modeling quality.
“Combine the exponential computing power with data connectors like Supermetrics and Funnel.io that can standardize the data formats and automate the data imports, and you get continuously updated MMM results with uncompromised modeling quality.”
End result: Incredibly fast data-to-results, automated updates once the model is configured, and unparalleled business impact/cost ratio. A true win-win-situation.
None of this would’ve been possible ten years ago, but it’s possible today.
And it will revolutionize how marketers can measure and optimize their marketing effectiveness.
Here’s how.
What are the benefits of continuous MMM
The key benefit of MMM is that you get extremely accurate validation for what’s working in your campaigns and what’s not, as well as implications on how to improve the effectiveness further on.
The key benefit of continuous MMM is that you’re in a constant test-and-learn loop that enables you to move from reactive to proactive marketing management:
Improved marketing effectiveness. As optimization is all about testing and learning, constant test-and-learn loops enable continuous optimization cycles
Faster fact-based decision-making. Always-up-to-date insights enable you to act on changes in marketing effectiveness well before others
More accurate forecasts. Continuous MMM enables the model to improve its forecasts with each update, making the forecasts more accurate over time
The difference between MMM projects and continuous MMM is like switching maps to a GPS: No matter how the circumstances change and develop over time, you’re always on the fastest route to success.
“The difference between MMM projects and continuous MMM is like switching maps to a GPS: No matter how the circumstances change and develop over time, you’re always on the fastest route to success.”
A well-known quote “You can’t manage what you can’t measure” by Peter Drucker encapsulates why measuring marketing effectiveness on a continuous basis provides a competitive advantage in today’s world.
The world we live in keeps on changing faster and faster as time goes by. To manage the change, you need to be able to measure it.
And when the change is continuous, so does the MMM need to be as well.