Attribution and Big Data: Why an Attribution Model Matter

Attribution Models

EMarketer recently discussed the conflict between the necessity and challenges surrounding attribution in their webinar, “Making Attribution Work in the Age of Big Data.” As data becomes more readily available and invaluable to advertisers, the tasks of sorting, classifying and applying that data correctly become vastly more important.

Attribution Model Importance and Challenges

When the necessary data can be collected and interpreted, attribution allows marketers to fine tune their campaigns to meet goals more efficiently and reach target audiences more effectively. However, proper attribution can be a challenge.

Consider: Many marketers currently only look at “last-click” attribution – or what was the last channel that led to the action (i.e. banner, CPC, email, social, etc.) However, the actual touchpoint that spurred action may have come earlier. If so, what aspect of that impression was most important? Were combined efforts responsible for the action? Did both online and offline campaigns play a part?

Understanding how all of these channels work together is extremely important, but it’s also extremely complex. Currently, there are 7 main attribution methods, including last-click attribution, broken into two main types of model.

Types of Attribution Models

The various methods for attribution are typically arranged into two main groups of models: one-touch and multi-touch:

  • One-touch models usually involve either the first or last touch someone had with a marketing channel before they converted. Their major benefit is that marketers know exactly which views or clicks led to the action. Yet, one touch models can dangerously devalue critical advertising channels that actually assisted with the conversion. Furthermore, a last click model can get you into misleading areas on fraud and viewability. Last touch models also reduce the importance of mobile channels.
  • Multi-touch models give attribution to all the different channels that led to the conversion, including upper-funnel activities like content marketing or social media that very seldom directly lead to clicks or views. The drawback to multi-touch attribution is that it requires a lot more data and modeling capacity. Nevertheless, marketers increasingly favor multi-touch attribution because it allows for analysis of the interplay between channels. eMarketer actually predicts that by 2017, 50% of all markets will use a multi-touch model.

Several models exist for assessing attribution. Unfortunately, no one model stands out above the rest. Choosing the right model to assess attribution for your campaign will depend on your goals.  

Which Attribution Models Should a Marketer Use?

While one-touch modeling gives a clear picture of how someone entered or exited the marketing funnel, multi-touch provides insights throughout the process. It also helps determine how all of the marketing channels are working together and provides a more cohesive picture. In deciding how to model attribution, marketers need to think about what kind of data they have and what they want to know.

Attribution modeling relies on “Big Data” collected throughout the conversion process. Big Data greatly affects the efficacy of these attribution models. We will review the pros and cons of Big Data as it relates to attribution models in a future post, so stay tuned!