How to successfully implement Robyn & Meridian

“We tested MMM internally and couldn’t capture all our digital channels accurately.”

Not the first time we’ve heard this.

Many times here at Linea, we talk to marketers after their data science team tested a free MMM tool (Robyn/Meridian). The model had been completed, but the results raised more questions than answers.

Three primary issues to overcome in “Free MMM”:

🎯 Mid-funnel undervalued (YouTube/CTV)
This was the core use case for MMM — yet mid-funnel impact was missed. Neither memory effects nor interaction effects (e.g. YouTube driving search) were properly measured.

🔍 Search overestimated
Generic search looked incredibly effective. However, high ROI often reflects demand capture, rather than true incrementality. We recommended revisiting the modelling structure to reflect what was actually driving search behaviour.

🤔 The team didn’t trust the results
Two possible reasons:
(1) A disconnect between data science and marketing. A third party can help bridge this gap by aligning both sides
(2) A belief that measurement is “one and done”. In reality, it should be treated as an evolving process

Free MMM tools can be a great starting point, but without the right expertise, they often lead to misinterpretation.

Accurate measurement needs people who understand both marketing context and modelling nuance.