Google Analytics Premium, the powerful fee-based version of Google Analytics that came out in September 2011, offers numerous advantages and functions.
Prior to the on-line release of this “Premium” version, with Google Analytics it was possible to define a pre-built or personalized attribution model: to analyze using multi-channel funnels (indirect conversion paths, time, length) and to analyze the contribution (conversion and objectives) of each channel, campaign, source, etc.
The goal of this new version is to offer greater functions especially from an Attribution point of view that I’ll summarize for you.
With Google Analytics Premium, in addition to standard functions, you can now:
1. Design exactly the same reports but across all your data not only a sample as compared to a classic GA. This function provides greater accuracy for sites with very high traffic.
2. Access to a dynamic attribution model, Data-Driven attribution, to evaluate thousands of combinations of channels in order to find synergies (based on the mathematical Shapley model). What’s new is designing an attribution model from a probabilistic model.
This Google Analytics Premium function has several aspects.
1. Data-Driven Attribution makes it possible to analyze conversion paths for each visitor to a site.
It is based on a predictive model of the channels’ impact on final conversion. It then attributes a value to each channel.
2. Then, it defines a conversion rate for each visitor coming from a specific conversion path by comparing it to another rate for a same group of visitors but with another conversion path.
3. You can use the “Model Explorer” report to compare the results and contrast them with your default attribution model to adjust it.
This new version of Google Analytics also offers a particularly interesting new function: the addition of a Post-View in the conversion channel analysis (Double Clicks Manager link required).
You can link DoucleClick Manager and Analytics Premium so the Post-View data appear within the conversion channels. This makes reading easier and allows you to take the post-view into account in your attribution analyses (e.g., direct access from opened emails).
The advantages for attribution are certainly limited nonetheless they are still interesting to exploit notably the “datadriven” which makes it possible to move from a Data-based model to a predictive model. The primary challenge is to improve the quality of your attribution models while allowing you to make quick and efficient decisions.
It should be noted that linking with DBM is a true advantage. However, we regret that the time between 2 keystrokes hasn’t been reduced (24-hour minimum).