In today’s truly digital world, a customer is exposed to multiple campaigns running via multiple channels. Determining the contribution of each channel and campaign finally leading to a conversion and attributing credit based on this is called as Multi Touch Attribution (MTA).
Much like a game of football if we just look at the first (kick off) and last (shot to goal) events we lose the intricate moments of how the goal is scored! It is of utmost importance to understand how all the various players are working together!
Traditionally, all the credit was attributed to the last impression/click which led to a conversion, also referred to as Last Touch attribution. Some have also attributed all the credit to the channel where the customer is first exposed to the marketing activity which is called as First Touch attribution. These two techniques however fail to capture the customer journey, the relationship between multiple touch points the customer might come across.
Rule-based models work on predefined conditions to attribute the credit. A brand might want to give equal credit to every channel which has served an impression to a customer. This can be achieved using a Linear model where every touch point is accorded equal credit. On the other hand, a brand who does not want to completely step away from traditional first and last touch models use a U-shaped model where first and last touches have a higher attribution of credit than other touches.
Rule-based methods are a great and simple way to attribute credit. Deciding the model however is done on either a hunch or based on heuristic results. With the extensive amounts of data present about digital spends, newer data driven models of doing Multi Touch Attribution have taken shape. Two leading data-driven techniques with wide industry adoption are Markov and Shapley models.
Cartesian leverages these methods to assign partial credit to all marketing touches involved with a conversion. This helps marketers address problems like getting an accurate picture of which marketing touches are working effectively and efficiently for them, justification of marketing spends on non-first and non-last touches leading to conversions and insights on reallocation of marketing budget.
Cartesian’s data-driven models work well with data from major web analytics platforms like Adobe Analytics and Google Analytics. Along with this, customization for acceptance of data from other analytics platforms and ad servers is also offered.
When insights from data-driven models are used to reallocate budget and measure lift with historical campaigns, the typical impact seen is in the ballpark of 25% reduction in CPA or 10% increase in ROAS.