Google recently launched the latest version of their "Modern Measurement Playbook" that talks about combining Marketing Mix Modeling, Attribution, and Incrementality testing.
What's great about the playbook
💡It openly discusses what I've been preaching about recently: Ad platform- and attribution -reported ROAS metrics do not reflect incrementality. For example, we often see Google Analytics 4 -reported ROAS for branded search be 2-10x compared to MMM-reported ROI.
💡It describes (on a high level) a method we've been using already: You can improve granularity of MMM insights with smart use of incrementality factors and ROAS data from Google Analytics 4 and Ad platforms.
💡It provides some practical tips and ideas for incrementality testing.
After reading the playbook, I realized that there’s nowadays more ideas available for building a high quality Marketing Mix Model than ever before. But it means that building an MMM solution is more complex than ever.
What this leads to
👉 Teams building and operating in-house MMMs will have higher and higher demands in terms of what is expected from them. Delivering a model that meets the standards of today will either mean building bigger data science / engineering teams, or betting that a few multitalents will be able to do it (and stay in the company to improve it over time).
👉 Consultants delivering one-off MMM projects will have increasingly hard time delivering high quality MMM results in line with today’s standards. It's because MMM's complexity has increased, but also because building a high quality MMM capability is a journey. For example, it is unlikely that an advertiser has incrementality tests in place across media on day one.
👉 Competition among MMM SaaS companies will intensify as there starts to be larger technological differences between solutions. I already see some companies embracing the latest ideas, and some sticking to the old.
It seems that it is more exciting than ever to work with Marketing Mix Modeling!