Jane Christian, head of analytics and insight, EssenceMediacom, believes an augmented version of Media Mix Modelling can achieve the deep and detailed insights into live digital campaigns that marketers need.
Multi-Touch Attribution (MTA) is broken. As we enter the swansong of the third-party cookie, the industry is finally admitting that not only will MTA no longer serve its function, but it never really did. MTA promised to give us regular and granular insights into digital performance, to help us optimise live in campaigns. However, the promised micro-digital measurement solution could only ever really claim to be better at attribution than last-click data – not a high bar, proving that anything centred on tracking and not incrementality is doomed to failure.
After years of the industry talking about leaving Media Mix Modelling (MMM) behind, we are seeing its resurgence. Just as 70s flares now once again roam the streets, MMM is back to being recognised in its rightful place as the gold standard for measurement.
However, the renaissance of MMM is not the end of the road – while it is the best tool we have in many ways, it still has its pitfalls. MMM is slow and often not granular enough. While it does measure incrementality, it fails to provide the granular insights modern marketing relies upon for live campaigns.
And so MMM alone is not the answer. Just as 70s flares are now often paired with a set of AirPods, so MMM must be paired with modern technologies. At EssenceMediacom, we are betting on what I’ll call MMMM: media mix modelling, and more.
There are (broadly) two ways to measure the incremental impact of media – modelling, and test and control. Modelling techniques such as MMM use regression analysis over time to disentangle the incremental impact of media from the other factors that drive sales, such as price, distribution and brand equity. Then test and control: advertise to one group and not another, and measure the difference in response between the groups.
To supplement and complement MMM, there are three innovations we find deliver that much-needed granularity and speed.
Automation of MMM data feeds: Our suite of products, known collectively as emRapidModelling, automate MMM data feeds and the time-consuming elements of the modelling. This allows us to deliver those insights on a monthly basis, meaning our analysts’ time is better spent on insight generation and embedding learning. This takes MMM from being something to help with strategic decisions a couple of times a year to something that provides insights every month to help with ongoing optimisation. We also use this to refine forecasting. Leveraging automated MMM feeds (to ensure more regular delivery of these insights) that also empower predictive intelligence is key in a modern media programme.
Granular MMM: Rather than modelling using aggregated, weekly data, we can model at both a daily and postcode level, producing incredibly granular media effectiveness learnings to help optimise within a channel – for example, at a tactical, creative or regional level. This can unlock a new level of optimisation learnings beyond what we’ve enjoyed from traditional MMM alone.
Geo experiments: There are plenty of exciting new solutions on the market that are fuelled by geo experimentation, providing incrementality at a granular level for digital channels. The principle is simple – to run continual experimentation at a city level by either dropping spend in a selection of cities or increasing spend. This way, we can get a very clean read on both the incremental value of each channel, plus how they perform at different spend levels.
Together, these methods level up speed and granularity, unlocking more optimisation learnings than MMM alone.
Ultimately, though, delivering this approach is about more than just the tools leveraged. The hunger for delivering measurement innovation needs to be embedded in the organisation. It’s not just about using the latest technologies, but also building a team with the right specialisms and making the most of the full potential of data. If these can all be built into a team, then we can as an industry truly tap into the value of ‘MMMM’.
Head of analytics and insight