Business | 13.03.2014

Cross platform attribution, or how I learned to stop worrying and love data

What is it?

Sounds complicated but it need not be. At its most basic, cross platform attribution is about understanding what factors influenced a consumer to buy something and in the context of marketing, what role did media play.

How do you do it?

There are two common methodologies used and in fact, it is possible to combine both into a hybrid approach:

  1. The Top Down Approach
    This approach looks at what media is being used to reach potential consumers. Many models also factor in variables that can influence purchase decisions, such as the price of goods, income levels and inflation. All of this data is then used to build a model that can tell you the role each form of media played in influencing a purchase. This can in turn be used to inform decisions on how much of each media should be used to maximize consumer purchasing.
  2. The Bottom Up Approach
    The Bottom Up approach takes advantage of the fact that today, particularly in the realm of digital marketing, we have access to very granular data at the consumer level.

The Outcome

If you can measure exactly what ad impressions each person was exposed to, when they were exposed and even where they were at the time, you can, in theory, build up a pathway that tracks each purchase back to an individual consumer. Over time, as data is collected, a statistical model can be built up, which will give us the optimum mix of media needed to convert a consumer into a purchasing customer.

While on the surface this seems simple enough, the difficulty in building accurate models lies in the collection of data and the quality of the data itself. For example, measurement of digital media typically employs the collection and use of cookie data, and we all know that consumers regularly delete their cookies.

This can and will skew a model that uses cookies to measure consumer behavior. Tracking exposure of media on mobile devices is another example of how it can be difficult to collect all the data needed to build an accurate model. The type of tracking that is possible when a user is on a web browser is often difficult to achieve on mobile devices.

Should mobile media play a significant role in the pathway to purchase of a particular product or service, then current data collection barriers would mean that our model wouldn’t be particularly accurate. Connecting online media exposure to offline purchase is another obstacle, one that may require organizations to implement operational changes at their point of retail.

These are all issues that can and most probably will be overcome through technological and organizational innovation. However, it is unrealistic to assume that it would be possible to build an attribution model that is 100% accurate, and it is not a necessity either. While your current data set may not give you this level of accuracy, it will still allow you to make better informed marketing decisions than those with no model at all.

Brands that begin to invest in such marketing tools today will find themselves ahead of the curve. This is particularly pertinent for those that may not be able to outspend their competitors but need to out maneuver them.

As a marketer, regardless of the budgets you may have at your disposal, knowing that you are spending your media efficiently and maximizing outcomes is most likely one of things that keeps you awake at night. Investing in attribution modeling is a significant step in right direction to achieving that goal.

Join the conversation

Christos Solomi
Group Director-Programmatic at RESOLUTION