Traffic Attribution Modeling in Google Analytics - Easier with Visualization

Google Analytics offers an easy entry into the world of customer journey analysisfor quite some time now with its Multi Channel Funnels reports. This year Google has made another big step and enhaced it with the Attribution Modeling Tool which allows us to create and compare various attribution models.

For online and performance marketing people this is extremely important and helpful - and complex at the same time.

With some simple visualization techniques it is much easier to gain insights - this article shows how.

Preparing the Attribution Models

The tool allows us to compare up to three models at the same time. In this example we want to examine the beginning, the middle part and the completion of the journeys. Therefore we choose

1. first interaction - a standard model
2. supporter - a custom model
3. last non-direct click - a standard model

The “supporter” should include all touchpoints after the first and before the last click. For this we create a new model based on the “position based” model and allocate 100% of conversion credits to middle interactions and zero to the others:

It makes a difference whether we choose “last interaction” or “last non-direct interaction” since in the latter case “directs” have less impact. Whatever implication this may have needs to be analysed in further comparisons.

Now this is the result we get in Google Analytics:

Note: In this case we work with a custom channel grouping - named “custom grouping”. This is another important strength of this tool. Here, for example, we chose to put all brand keywords in their own channel “brand search” because we believe that those searches are very different compared to generic searches and should not be included in the standard search channels (organic and paid).

The figures show it all - don’t they?

We can definitely gain some insight out of this with some experience and time at hand. Also the delta columns on the right help, specially the arrows.

But does everyone really get what we can see here? And more importantly: what can we deduce from it?

Frankly, I’m not sure. This is why I click on the small grey icon in the top center (just beneath the “Conversions & Value” dropdown menu). Now the table changes as follows:

The color gradient immediately shows which channel

• introduces the contacts,
• is supporting
• or closes the deal.

In our example, we see in lines 5 and 6 that e-mail and organic search (without brand keywords) are not introductory channels but have a supporting and finalising effect. Lines 1-3 show in this case a strength throughout all brand channels supported by paid search.

What can I deduce from the attribution models

So what? These initial findings are a good analysis of the current situation. As a next step, however, we can go into more detail by means of drill-downs.
What search campaigns do help, in which phase and to what extend? Which channels are initially important so that someone will close the deal via the brand search in the end? If I change the email frequency, what happens then? (To name just a few questions ...)

This specific example does not only show what insight we can potentially gain from it but also the complexity when we want to derive new strategies: It won’t be possible to stop the clear focus on brand as shown above from one day to the other - but it will be possible to tackle it in a more targeted and controlled way.

I want that too!

Great - here you go:

To do this yourself, our free extension for Google's Chrome browser, namedTable Booster, is sufficient. It can be downloaded from the Chrome Store here. All of the features this tool offers when using Google Analytics, can be found here (there are lots more!). Please rate the plugin in the Chrome Store - many thanks!

If we can help you in analyzing your customer journeys: please contact us - this is our (exciting) job :-)

And if you liked this article, then it would be great if you share it: