When we look into Analytics and Adwords, we know that Google chooses some settings for you. Most of the time in Adwords we move away from these so-called recommended options, but why do we not do the same with Analytics?
The key and singular most important piece of data is arguably conversions. So why is there discrepancy between Adwords and Analytics for the same conversion? The answer is attribution modeling. You may think Analytics/Adwords work off last click/interaction modeling. Unfortunately this is not true. Analytics work off last non-direct click and Adwords work of last Adwords click. This means without knowing it, most marketing is done with misinformation.
Well, good news data lovers, there is a setting that moves away from last click attribution and move onto almost any type of weighting you wish. But donâ€™t believe me before you take attribution modeling as gospel, inside analytics -> conversions there is a sub choice of attribution model comparison. Inside this very powerful tool is the ability to dive into your data and change the attribution, at will, on existing data.
Below I have taken 3 different attribution models. Last non-direct click (the normal Analytics view), position based (my favourite view) and last Adwords click (the Adwords view). You can see the massive difference in percentages between each view.
But better yet, below you will see, after moving over to the Adwords view on this tool, most campaigns are again showing a massive percentage difference. This means that all decisions would wildly change depending on which attribution model you are looking at. Â We can see that the generic campaign in position-based attribution is out performing the best last non-direct click.
Do not despair there is a solution! Inside admin, under view, you will see a choice called attribution models and you can choose another option. What option should you choose though?
As I said before the three main choices are:
Last interaction: we touched on the inadequacies of missing data already, but there is some importance of having one view that is last interaction. Every other tool uses this attribution weighting; therefore to see what everyone else is seeing, you need at least one view that uses this model.
Linear: This takes every single interaction and weights it completely evenly and gives you the most all encompassing view of all the data. A good attribution to use to get a clear idea of what are the most touch points in an account.
Position based: This is my personal favourite view, as all data is not the most important. This heavily weights the first and last click of the conversion while not ignoring the middle interactions. Taking the first interaction, where the customer finds the account and the last interaction, where they actually buy, as the most important.
Below you can see the difference between all three on the same account.
Finally there is an option inside analytics to custom model. Â So let’s do this by creating my favourite custom model. Open up a new view (we donâ€™t want to mess with already collated data). Jump into attribution models and select position-based model. Give it a name â€ścustom model 1â€ť . Then change first interaction down, it is important but not that important (remember this maybe different for every business and one model may not fit all). Let’s have a weighting of 30, 25, 45.
We want the first interaction to be a little more important than middle and last interaction the most important. Next let’s open lookback window, I suggest changing this depending on how many touch points it normally takes to convert, as it will only count what is important. For this business we know that it takes a while to make a decision, but no longer than a month; so I will set it to 30 days. Â
Activate adjust credit based on user engagement and change it to page depth (I prefer this over time on page, because it means that they have to be their rather than making tea. This one works on 1 page websites). Open up custom credit rule and choose interaction type exactly matching click and give it credit of 1.2. This will weight Adwords clicks up by 20%, highlighting the data we are interested to see.
It should look something like this.:
Apply it and enjoy your newfound data.
Remember the power, of attribution models is that it can change depending on the business. If the window of decision making is shorter change the lookback window down, or if the main KPI is getting the brand name out there you may want to take a look at heavily weighting first interaction and opening up the look back window. If prospects come from a specific source where lead quality is high Â then in custom credit rules you can weight that source to be 200% stronger. Have fun, it is a powerful tool to master!