October 25th, 2017
At Branch, we have been hard at work on a complete overhaul of our analytics systems. We process and analyze billions of events per day, and we have learned quite a bit about scaling systems in the process. Our Dashboard has been through several evolutions on account of scale. We’ve also heard and responded to your need for advanced analytics such as predefined events and different attribution windows. Given all that, we’ve invested the last 10 months re-writing our analytics platform to help us scale for a long time while providing you with enhanced analytics.
Based off of your feedback over the past 3 years, we set three main goals in revamping our analytics platform.
We’re excited to share that you can now try the new analytics platform! Starting today, October 26th, you’ll see a banner on the summary page of your Dashboard as shown below. It will allow you to switch into the new Analytics environment:
If you switch, you will notice a change in your install, click, and event numbers, all of which are based on the new analytics system. You’ll have the ability to preview the new Analytics platform, and switch back and forth freely between the new and old platforms, at least until early 2018.
Now, let’s explain why your numbers look different on the new platform. It all starts with our first goal of expanding attribution beyond deep linking. With deep linking, we matched users based on a configurable 2-hour window. We chose this length because any user opening a new app for the first time after more than two hours is likely not expecting a deep linked experience.
However, many of our partners use Branch not just for deep linking, but also for attribution. Branch is absolutely a useful tool in answering the question of “Where are my users coming from?” In the context of app install attribution, a 2-hour time window is far too narrow. If someone engages with an app install link today and installs the app tomorrow, you’ll want to have this information in your analytics platform.
As a result, Branch now has separate mechanisms for deep linking and attribution. If a user clicks a link and installs the app a day later, by default her install will be attributed back to the initial click, but she will not be deep linked. When you try the new analytics experience, you’ll notice the charts numbers reflect this model of attribution. Practically speaking, this means you will see an increase in the number of installs and opens tracked for any given campaign or channel.
Speaking of counting installs, we are now leveraging a more extensive attribution model to help attribute more clearly and accurately. The problem with today’s attribution is that when your customers click your links on Facebook, Twitter, Safari, and Chrome, it’s likely that this person is attributed 4 separate times, because those are four different cookies for the same device. Branch’s new underlying attribution model can intelligently pick up on the same person in four different platforms, and identify them correctly, leading to a more accurate picture of your cross-platform attributions.
Finally, in addition to separating our attribution from deep linking, we unified our data models. Previously, our data model was simple: we tracked clicks and events. With the new system, we went deeper and started pre-defining more events in the Branch system. These include pageviews, Branch CTA views, SMS sends, and more. This standardization will help improve data quality, whenever you export data out from Branch.
We’ve gone into depth in designing and building the new analytics platform. For more details on the full list of events tracked, as well as all the changes discussed in this blog post, visit our documentation here.
We’d love any feedback you may have. Go on and give our new analytics platform a spin, and let us know what you think either by reaching out to your Branch account manager, or writing into firstname.lastname@example.org.