Understanding attribution is sort of like trying to sit down and watch all three “Matrix” movies at once. At first it sounds great, and some of it even makes sense. You get
very excited simply talking about it. Once you go deeper and deeper into the topic, it starts to unravel and lose sight of where you started. As a matter of fact, attribution is a very
important concept, and one that almost nobody fully understands. Sound familiar?
As Morpheus would say, marketing is in a shift toward technology for delivering efficiency and
performance at scale. Attribution increases in importance because it becomes the fundamental component of understanding whether your marketing is working, but for it to take hold it has to
become standardized and simplified. You might need a Ph.D. to develop an attribution algorithm, but you shouldn’t have to have one to implement it in your marketing practice. After
all, how many marketers do you know whom you refer to as doctor?
advertisement
advertisement
Let’s dissect what we know about attribution. As it was recently, and most simply stated to me, attribution has
three models.
First there is a model of “last click,” which means exactly what you would think: the last click prior to a conversion is given the credit. Obviously this
is a flawed model because it doesn’t provide any value to frequency, and frequency is important when examining a considered purchase. Additionally, it doesn't take into account multiple
channels and focuses solely on digital, which is myopic at best. No true marketer with a degree of cross-channel responsibility should ever use last click as a model.
The second model is
“media mix,” which takes into account multiple channels, past spending and past performance as well as current spend and performance. This is referred to as a
“top-down” approach because it looks at the past to determine the future. It’s scalable and highly accurate, but doesn’t deliver information in real-time, thereby
creating an accurate, yet inflexible and delayed understanding of your marketing. It is good for projections, though, and for deciding how to spend your money.
The third model is
“digital attribution,” a rules-based model in which value is assigned, based on a previously agreed-to system, for every actionable engagement or interaction of digital media. This
model is much stronger, but it doesn’t take into account other channels, like TV and print. Once again, this is a stronger model, but myopic for true cross-channel marketers.
Where the
truth would lie, as it always does, is somewhere in the middle. The model of merging media mix and digital attribution is where a strong attribution story resides. As more media
integrates technology, more media can be tracked using digital methods. With more technology in place, we can gather a more comprehensive real-time evaluation of performance -- and if we marry
that data with past spend and performance, we can start to identify trends and understand the interdependence of media. It’s this interdependence that’s most valuable. The
issue with attribution is the silo’d approach. If you eliminate silos and embrace the interdependencies, then attribution can be significantly more easy to understand than the second two
“Matrix” movies.
The landscape of attribution partners is getting more cluttered, so one of the things to examine when selecting a partner is what model they use, and who has
used that model in the past. Understanding your vertical and the ways consumers interact with brands in your vertical can go a long way toward determining if a certain model will work for
you.
Algorithms and experience: these are the red pill and the blue pill required for understanding attribution, and in this case you need to take them both.