When MMM and lift disagree

Marketing mix models summarise long, noisy spend against outcomes. A lift test asks what happened when you withheld treatment on purpose. They will disagree. That is not a scandal. It is two instruments pointed at different sentences.

The week they diverge, we ask three questions in order. What conversion and window did the lift test actually randomise? Which channels did the MMM even see, and at what granularity? Which stakeholders needed a number by Friday regardless? The third question is not cynical. It tells you whether you are doing measurement or theatre.

If the lift test was contaminated, believe the MMM only as far as its own caveats allow, and fix the holdout. If the lift test was clean and short, do not let a national MMM veto it because brand search moved in the same month. If both are messy, say so. Conversion lift measurement includes the right to withhold a confident number.

Lift Studio’s seventh module is a protocol, not a winner’s podium. Bring both outputs. Leave with a memo that finance can mark up, not a truce that says “insights aligned.”