Our musings on making video personal
Personalization is a key component of modern marketing. Here at SundaySky, we built our SmartVideo Platform on the premise that video is an extremely efficient vehicle in driving business results, through the delivery of the right message, to the right individual, at the right place, at the right moment. We have seen this promise realized across paid and owned media programs, with the likes of Comcast, AT&T, Atlantis and many other leading brands.
As we begin to speak to consumers with personalized messages, we must recognize that they should be treated uniquely as individuals when it comes to marketing measurement as well. For example, consider how personalization impacted the evolution of marketing attribution over the years. In the beginning, there was last-click attribution. As its name suggests, regardless of how many times a brand interacted with a customer, regardless of through which channels those interactions took place, there was only one interaction to rule them all: the last one. One-hundred percent of the value of – and therefore, credit for – the eventual conversion was assigned to that last click…which left exactly zero to all the rest.
There are some clear advantages to using this approach: it’s straightforward, very consistent, and easy to grasp and explain to higher-ups and others in the company who are not well-versed in the intricacies of the marketing mix. These pros are valued so highly by many, that despite its obvious disadvantages, a significant portion of organizations are still using last-click attribution.
The drawbacks of last-click attribution revolve around accuracy. By only taking into account clicks, you undervalue other forms of engagement, like view-throughs. By not crediting any other interaction on the path to conversion, you are biased against upper-funnel channels; moreover, even if your entire path to conversion included multiple interactions within the same “closer” channel like paid search, not giving credit to any other keywords the user searched for except the last one means you will suffer from a similar bias. Your budget allocation (both inter- and intra-channel) will be skewed, and therefore sub-optimal. In other words: garbage in, garbage out.
Over the years, attempts have been made to tackle some of these challenges using other attribution models: linear, giving equal credit to each interaction; time decay, assigning the most credit to the last interaction, and less and less to the others preceding it; or U-shaped, assigning the most credit to the first and last interaction, and splitting the rest evenly among interactions in the middle. Though these multi-touch attribution models get around the issue of putting all your conversion eggs in one basket, they still all suffer from the same flaw: they are normative. They suggest that marketers impose a certain point of view that defines what the world should look like, instead of having the data tell marketers what the world really is like. Not very data-driven, is it?
Enter advanced attribution.
Advanced (or dynamic) attribution models are predicated on the assumption that there is no one-size-fits-all attribution. Since every consumer is unique, so is the attribution model that accurately describes the true value that each interaction and each channel provides along the consumer’s path to conversion. Technology needs to be harnessed in creating, as well as maintaining, these models because cross-channel data, including the time that elapsed between interactions, repeat conversions, view-through vs. click-through, and dozens of other considerations need to be taken into account. The model is tweaked and honed as new data continuously comes in. And since no two brands will ever have the same data, no two advanced attribution models will be the same. Advanced attribution could even be termed “personalized attribution”: it attributes the right value to each interaction for each consumer, providing marketers with an accurate understanding of their customer base as individuals so they can more effectively engage each in a targeted, personalized way.
Adopting advanced attribution is not without challenges. You need integrations that will pour data from all marketing channels into a single location so that the model can be (relatively) easily built. You need to have the political discussions around making the switch, as people’s scorecards may look much different than they did before without any change on their part. You need to educate other stakeholders on a much more elaborate way to assign credit. You need to get executive buy-in for a model that constantly changes. Clearly, not an easy feat. However, if you care about optimizing marketing performance, and assuming that a personalized model more accurately describes your reality, then you are all but guaranteed to see a top-line uplift. That is, if you indeed adjust your budget allocation accordingly.
As personalized video gains momentum as a primary engagement medium between a brand and an individual, advanced attribution platforms will continue to evolve and take into account factors like the length of an interaction (much higher for video than for other mediums) when assigning credit. It is an exciting time to be in this space.