What does the optimal use of a channel look like?
What is the impact if the investment in marketing decreases (or increases) by 20%?
How much influence does a channel have on sales/conversion?
What is the best way to distribute my budget across different channels?
In addition, the model also includes non-marketing expressions. These are influences of, for example, season, promotions and price. You could say that a Marketing Mix Model gives you a helicopter view of your entire marketing. It is a rock-solid solution to give you real insights that you can use. A well-designed and fed MMM is able to give good predictions of the impact of your marketing. Ideal for scenario planning, building a solid marketing strategy and budgeting.
A Marketing Mix Model for every marketing department and CMO?
Although an MMM sounds like armenia phone number library the holy grail for marketing departments and CMOs, there are important considerations. For example, you need data. Lots of data over a longer period. And good quality data. Most Marketing Mix Models, both open source and closed source, require at least 2 years of good quality data to be able to predict. For many organizations, bringing together consistent data is difficult, especially over a long period.
Channels like Google Ads, Meta Ads and TikTok generate a lot of data at a high speed, such as reach, frequency and clicks. There are many data points that need to be formatted for the MMM and from which you need to select the right, relevant variables. The important thing is that you compare apples with apples.
Especially when you also use indirect channels, such as TV advertisements, you will have to take this into account in the model.
Large amount of data is needed source: Gorodenkoff / Shutterstock
Open source or closed source Marketing Mix Model?
The big tech guys have different Marketing Mix Models that you can use. For example, Meta has the model called Robyn , which is supported by AI/ML. Early March, Google also followed with the model Meridian . Both models are open source and use data libraries from their parent companies. The great thing about these models is that they are adaptable to the wishes of the organization and their goals. Although this is the biggest advantage, adapting and maintaining them also requires a lot of specialist knowledge from a data scientist. Not every organization has that knowledge in-house or available.
