It was a simpler time, with a more open web and rampant stacked and piggybacked cookies to allow for easier user tracking across ad exposures. Even then, many of us will also remember, MTA didn’t often work. Long and complex deployment windows preceded long read-out windows, with anticlimactic culmination around uncertain and often unactionable “insights.” The promise was great, but the performance was not.
Then those MTA independents were swallowed up by tech and measurement giants. They became part of monolithic (and monopolistic) tech stacks and lost all their agility to deal with a less open (more secure) web, the mobile shift, signal loss, and the inevitable rise of walled gardens.
Yet some people still pursue the MTA dreams of their childhood. It would be so cool if it actually worked! Every day forward will be even harder to fulfill the promises of MTA than the days past. Its time to let go.
Marketing Mix Modeling is still great, so keep that to move the big rocks around. For those smaller rocks, one bet is on a shift to what we’re calling MEA, or Multi-Experiment Attribution.
Multi-Experiment Attribution is enabled by the world of precision targeting we live in across most marketing channels today. MEA comes to life with test and control deployment through the use of microtargeted geographies, 1st party data activations, or panel-based marketing activations. Experimental design allows for a better understanding of what really matters – incremental lift. Does that thing you did incrementally drive the business?
Running multiple experiments at one time is much more possible today than it was in the past, when we lived in a broadcast world and needed to cobble “matched” markets together and stare at them for 12 weeks in a row to see if we could get a readable result. Today’s tech, data, and precision marketing tools make things easier with the right marketing science deployed in tandem.
MEA, over time, leads to a solid foundation for understanding the levers of marketing performance. Even better, when used in combination with mix modeling, it offers the ability to validate or calibrate those models accordingly for scenario planning and simulation.
The best news for those of us beating our heads against the MTA wall for over a decade, is that MEA is a future-proof, privacy-compliant alternative to understanding the incremental lift of different channels, channel combinations, and tactics within those channels. Really, that’s what matters most in the never-ending quest for better performance.
