Questioning Attribution Models

By Blake Smith

With different channels of digital marketing all aiming to drive traffic and, ultimately, conversions, how does one decide which channel truly deserves the credit for a conversion; to attribute or not to attribute, that is the question.

There are about 6 default attribution models that can be utilized to determine how credit for conversions can be attributed to touchpoints in conversion paths.

Let’s take a minor, formal and informal look at the 6 attribution models available in Analytics:

Last interaction model:

The last channel that the customer interacts with before converting will be attributed 100% of the conversion value.

Let’s ignore the beginning, middle, and close-to-end channels of the conversion path, and let’s give ALL the credit to that final channel that basically just reaped the benefits of the other channels’ hard work.

Last Non-Direct Click model: 

Direct traffic is ignored and the final channel that was clicked through by the consumer, before converting, will be attributed 100% of the conversion value.

Direct traffic is consumers who have been obtained by a different channel, so let’s act like direct traffic doesn’t exist and attribute 100% of the conversion value to the most recent channel, prior to the direct traffic channel. Again, ignoring the efforts of the beginning channels in the conversion path.

Last AdWords Click model:

100% of the conversion value is attributed to the most recent AdWords ad click before a conversion was attained.

Nothing else matters but the most recent AdWords click before converting, end of story!

First Interaction model:

 100% of the conversion value will be attributed to the first channel of consumer interaction.

Let’s reverse the Last Interaction model’s concept and ignore the middle and end channels of the conversion path, and let’s give ALL the credit to that first initial channel that basically just reaped the benefits of the latter channels’ hard work.

Linear model:

The conversion value is equally divided amongst all touchpoints in the conversion path.

The conversion path was a true team effort and let’s give everyone equal credit for that final conversion.

Time Decay model:

Exponential decay is the foundational concept behind this model. With a default half-life of 7 days, any touchpoints that occurred 7 days before a conversion will be attributed ½ the credit of a touchpoint that befell on the same day as the conversion. The exponential decay concept continues as far back as your lookback window allows.

We won’t completely ignore the efforts of the channels in the early and middle stages of the conversion path, but the closer the touchpoint is to the actual time of conversion, the higher the % of the conversion value that will be credited to that channel.

Position Based model:

A fusion of the First and Last Interaction models that divides credit between the first and last interactions in the conversion path, with the middle interactions getting a small percentage of the credit.

The initial and final touchpoints are the roots to this conversion, so let’s split a high percentage of the credit between these 2, and divide minimal credit to the middle interactions.

For more insight as to how to attribute the value of your conversions, sign up for a free digital marketing audit!