Attribution in Google Analytics

There is an important Google Analytics feature that helps us understand how multiple channels often work together to drive the macro conversions, in the process of collecting and going through multiple micro conversions, or little steps in the customer journey. 

 Now, more about default attribution: in Analytics there are many different ways to assign this:

1) Last interaction: attributes 100% of the value of given conversion in the last channel. Imagine that the user first clicked a search ad, than a display one but finally converted from another one in mobile. In this model all the value will be attributed for the last channel.

2) Last non-direct click: only good if your funnel is NOT a multi-channel one. 

3) Last Adwords click: will attribute the conversion, as you may expect, on the last ad in Adwords. Can be an option if you are evaluating different campaigns and want to chose the most effective one, since it ignores the additional data in the journey.

4) First interation: you will see that 100% is attributed on the very first click. I personally don't recommend this setup, unless you're new to Adwords (to give an example) and need some quick and potentially useful insights. 

5) Linear: will take into account each and every channel in the cycle. 

6) Time decay: Good one with timely offers. This is based on exponential decay (sounds tough, but you'll soon realise that is not like that at all) - the model has a lifetime of 30 days and will measure in a way the touchpoints closest to conversion giving them different credit (1/2 if less than 7 days, 1/4 if less than 14, etc.).  
7) Positions: which is a mix of last and first interaction. Helps you to split the percentage of credit between them. Although the most common one is 40-40-20, I recommend setting it up to 30-30-40. I will explain later on why. 

Once we set up the attibution models, we can compare them in GA in Model Comparison Tool. Awesome, isn't it? It's as easy as setting up a model (let's say, first interaction first) to see the Conversion Value and then set up other models to compare the results.

The key is in experiments, and those are going to depend completely on your business. Think about it: if what you sell is a Master Degree or MBA, for example, for a top tier business school, your goal is to generate leads. The path towards quality leads is long and complicated though, since you need to hit the right user (with a high probability of having all the requirements met in order to be admitted) and those are going to go through a long journey in order to verify what school they want and what's worth their money. Attributing 100% to first click in this case would be unwise, since the average visits and navigation not only though your site but also different channels is going to be much longer. What is an average then in time and clicks? It depends, but the users may spend as much as 6 months and as many as 250 hours navigating in order to make this decisions. You want to know which blog entry is the strongest one towards the conversion and which are the sites the users visit before it (if its Finantial Times, you may want to boost your rankings ;)).

The thing is going to be different if you have an e-commerce that sells cards or small presents. The average time spend on a decision making process in which card to use is going to me much shorter (VERY much) and you can easily attribute 50% on your last interaction towards the conversion. 

The best thing you can do once you have insights and tested your business is a custom attribution model. This is a specific model for your and only your business (how great is this?). To do this quicly in your GA, you Create a new custom model, enter the name, use the baseline as a starting point (don't worry, this is only a starting point based on which the rules will be applied), save an apply. That's it!