Google has recently released the latest version of their ‘Modern Measurement Playbook’ describing their views and recommendations on how to set up a modern measurement framework. They describe how this framework should be constructed of three elements: Attribution (MTA), Media Mix Modeling (MMM) and Incrementality Experiments. At ScanmarQED we embrace triangulation and have discussed the practical aspects of such a framework in our recent webinar.
What we like about the playbook
⇒ It prescribes a comprehensive overall framework.⇒ It highlights where each methodology shines:
- MTA is not dead - data-driven attribution is here to stay and will retain a key role in the measurement toolkit. Of the three methodologies, MTA is the only metric that works in real time.
- MMM is enjoying a rebirth.
- Experiments are an integral part of triangulated approach.
Our thoughts on implementation
⇒ Google describes the playbook as a "media” measurement framework, but we’d like totake that one step further and call it a marketing measurement framework.
⇒ Google describes MMM as a twice per year activity, but shouldn’t the frequency of
measurement depend on the frequency of decision-making? We believe that with the right
data management system and model update algorithms, it is possible to deliver many
more frequent model updates. We empower clients/teams to run MMM as often as is
helpful.
⇒ With triangulation, you can start building up your framework gradually – first with MMM
only, then add MTA, etc.
⇒ In many cases, 2 out of 3 approaches may be sufficient:
- If you don’t do lots of offline spend and your sales is online, you can do experiments plus MTA.
- Using a designed experiment to give all MTA roas a haircut is a useful guesstimate, but the same can be done with MMM.
- If all digital placements are being managed by auto test and learn processes in CM360 or in the Meta product that does that, then MTA can be dropped and MMM and lift studies can cover all measurement needs.
A few points to be aware of
Experiments:
⇒ Geo experiments might work well in the US, but can be harder in smaller countries whereit isn’t so easy to find comparable geographies in which you can control exposures.
⇒ Experiments are harder to perform across multiple platforms (e.g. Google vs Meta).
⇒ Easy to measure incrementality, but harder to measure curves (diminishing returns).
MTA:
⇒ Be aware of the limitations of GA4.⇒ We recommend using an independent MTA solution that can fully account for upper-funnel
activities across other walled garden platforms.
Non-Google platforms:
Clients need to carefully consider how to set up their measurement framework to cover all channels. Open-source MMM vs strataQED
By leveraging a triangulated approach and by addressing the limitations we’ve outlined here, marketers can unlock the full potential of media measurement to achieve their strategic objectives. This layered approach can be difficult to implement as there is no one-size-fits all measurement solution; ScanmarQED is here to help architect the best fit for each business.
