Attribution Part 4 - Single Source Attribution Models

When I was a kid in the U.S. there was a rhyme about how to calculate your income tax.

“How much did you make?
How did you spend?
What have you got left?
Send it all in!”

Definitely not what you want from a tax system but calculating the return on your marketing investment should be this straightforward. Unfortunately, it’s not.

In fact, there are many different ways to attribute revenue to your marketing activities, but attribution models can generally be put into three buckets. In this post we will look at single source models. We will look at multi-source models in the next and then unified marketing measurement after that.

Single source attribution models as their name suggests match sales revenue to only one marketing tactic so one expense. They are very easy to implement, easy to use and easy to explain. 

1.     First Action attribution gives all of the credit of the sale to the channel or tactic which results in a customer’s first interaction with the brand. Sometimes called first touch, this model is great when you want to assess which channel brings you the best leads.

2.     Last Action attribution (not surprisingly) gives all of the credit of the sale to the last action or last touch. This model is great for assessing which channels drive sales.

Unfortunately, this simplicity will lead to bad decisions. The rest of the journey is not tracked so the contribution of those other touch points is unknown and spending on them is impossible to justify.

An example from the airline industry shows how looking at only one data point can lead to bad decision making. In the early 2000’s, an airline surveyed passengers about their onboard experience and a common theme emerged about the lack of legroom. In response, they looked for ways to increase what is called seat pitch or the distance between seats. It turned out this would have a significant impact on the airline both in the cost of reconfiguring the planes and lost revenue.

Luckily before they went to all that trouble and expense, one of their business analysts noticed a correlation between complaints about legroom and the number of minutes the flight was delayed getting to the gate. It turned out that a late arrival of as little as ten minutes caused a steep rise in legroom complaints. Instead of reconfiguring all of their planes, the airline just adjusted their schedule so that flight times increased by 15 minutes. By just looking at the complaints about seat pitch, the airline would have made a very costly decision.

This is probably a good time to mention that all of the models mentioned in this and the next two posts have a place in your marketing tool kit, but you need to understand their strengths and limitations and use them accordingly. Check out Part 5 to find out about the models which look at more than one touch.