How to successfully implement your MMM programme?

Half of our customers have tried open-source Marketing mix Modelling tools before using Linea.

So why do they come to us?

Firstly, these open source packages (think Robyn/ Meridian/ Python packages) solve a very specific issue: they allow you to run a model to understand what's driving your business.

But the process of running an MMM model is bigger than that. To successfully implement an MMM programme, you need to solve three things.

X Data set up
✅Run model & analysis
X Take action from measurement

With open source models, solving the “running a model” part. Many MMM providers, like ourselves at Linea, are set up to ensure that a brand can solve all of the areas, including data ingestion & taking action from measurement.

The questions that you need to answer to set up an MMM programme:

➡️ You need to connect and clean your data

Setting up an MMM programme isn’t a one-time thing. You need to ensure that you are getting timely data transfers. We recommend that ~80% of data should run automatically into your data pipeline via APIs, with ~20% requiring more manual interventions

➡️ Create a process to select the right model

What is the goal of your model: Fit? Predictive accuracy? At Linea, we offer support to review open-source built models. We offer a five-step process to review: Modelling structure, missing variables, tests & transformations, benchmark & key questions you could answer.

Noticeably, half of those are focused on modelling quality & half on setting up your measurement to take action.

➡️ Translate outputs into insights marketers can use

➡️ Build trust across marketing & finance teams

This area is to focus on building trust. The specifics at organisations will differ, but the important aspects are:

• Sharing the approach transparently
• Framing an MMM measurement alongside other measurements
• Taking action early

➡️ Take action from measurement

This is the most important component. No point building an MMM programme (internally or externally) if you aren’t going to take action. We have seen that many teams struggle with this part.

Focusing on the key questions you set out to answer is important. Was it to see if you could scale your Facebook spend? Was it to allocate budgets? Make sure you execute against those questions. At Linea, we have noticed that many teams struggle with this phase; that's why we built our Linea Scenario Tool to help businesses optimise their marketing activity. Try with a free trail today.

That’s where open-source tools fall short. We hear it all the time:

• Marketing teams don’t trust the outputs they’re given
• Data science teams build models, but can’t keep up with the speed of marketing needs or turn results into actionable insights

Running an open source model can be easy, but taking action is the hard bit. So if you want to run MMM, you need to make sure it is set up to succeed & that's why they come to partners like us.