The campaign has been running for a month now, you’ve spent 10k trying to get people to buy seats for a conference your company is organizing, you did your homework and selected multiple digital channels according to your target (Social, Display, Search, Referral, Email.) In order to know the success of your campaign and optimize for future ones, your boss ask you to present a detail ROI, explaining how much of the budget you invest in each channel and how many sales can be attributed to each of them.
This is called in marketing “Attribution”; the process of pointing down which actions or events contribute to achieving a desired conversion (sale) and it which proportion (credit). The concept of attribution started to gain traction when marketers begun to implement campaigns across multiple channels, the need for measuring the effectiveness and contribution of each channel to the overall campaign emerged, but the challenge was how to do it? You could attribute a percentage of the sale to a TV ad and another to a print ad, but how did you account for word of mouth? or for brand value?.
Of course this being a digital marketing blog, everything starts with the internet, with the use of this new channel by marketers, new tracking and accountability capabilities were now at their disposal, with a little bit of statistical knowledge this data could be used more effectively towards measuring attribution of campaigns.
Avinash Kaushik is one of the most respected web analytics experts in the industry, being part of the original Google team and the writer of several books about the topic he certainly has the street credit to talk about attribution. In his blog Occam’s Razor, Kaushik presents us with the challenge of measuring attribution in digital channels. He also provides us with several multi-channel attribution models that you can use (while in Google Analytics):
- Last Click Attribution Model: this model attributes the 100% of the conversion to the last interaction made by the user. For example; a user came to the amazon website to buy the product that he left on his shopping cart last time he was there, an email remind him that he had this product on his shopping cart, he clicked through the link of the email and finished the purchase. In this case 100% of the sale is attributed to the email. There are obvious disadvantages in using this model, hardly a conversion can be attributed to only one thing! but in my opinion is better than not measuring attribution at all.
- Last Non-Direct Click Attribution Model: this model is very similar to the first one but it discriminates the direct interaction that often occurs before a conversion. Most of the time the conversion end up being attributed in a 100% to the referral (place from where the user is coming).
- First Click Attribution Model: rather than giving 100% of the attribution to the last click made by the user, this models reverts it and focus on the first interaction done by the user, most of the time can be tracked backed to the search or nowadays to social.
- Linear Attribution Model: in this model the marketer tracks back all the interaction (steps) made by the user before converting and assigns equal credit to each one them for the convention. Its more comprehensive than the models we saw before because at least gives credits to other actions in the funnel.
- Time Decay Attribution Model: It parts from the premise that the touch point that interacts the latest before the conversion gets more credit than its predecessors.
- Position Based Attribution Model: This model parts from the premise that the first and last interactions are the more important ones, thus assigning more credits to these ones and distributing the rest among the other touch points in the conversion funnel.
- Customized Model: Using another model as your starting point and then layering other factors on top of it to reach a model that best fits your conversion funnel.
If you are not measuring attributions of your online campaign, you should start now!, the Time Decay model is a great starting point, just remember that every touch point influences the conversion and that the funnel may look very different depending on what you are trying to accomplish.
For more information on attribution model or a detail guide of how to implement them in Google Analytics I invite you to check out Occam’s Razor: