Setting After The Fact Metrics in Adobe Analytics

As loyal blog readers will know, I am a big fan of identifying business requirements for Adobe Analytics implementations. I think that working with your stakeholders before your implementation (or re-implementation!) to understand what types of questions they want to answer helps you focus your efforts on the most important items and can reduce unnecessary implementation. However, I am also a realist and acknowledge that there will always be times where you miss stuff. In those cases, you can set a new metric after the fact for the thing you missed, but what about the data from the last few years? It would be ideal if you could create a metric today that would be retroactive such that it shows you data from the past.

This ability to set a metric “after the fact” is very common in other areas of analytics and there are even vendors like Heap, SnowPlow and Mixpanel that allow you to capture virtually everything and then set up metrics/goals afterwards. These tools capture raw data, let you model it as you see fit and change your mind on definitions whenever you want. For example, in Heap you can collect data and then one day decide that something you have been collecting for years should be a KPI and assign it a name. This provides a ton of flexibility. I believe that tools like Heap and SnowPlow are quite a bit different than Adobe Analytics and that each tool has its strengths, but for those who have made a long-term investment in Adobe Analytics, I wanted to share how you can have some of the Heap-like functionality in Adobe Analytics in case you ever need to assign metrics after the fact. This by no means is meant to discount the cool stuff that Heap or SnowpPlow are doing, but rather, just showing how this one cool feature of theirs can be mimicked in Adobe Analytics if needed.

After The Fact Metrics

To illustrate this concept, let’s imagine that I completely forgot to set a success event in Adobe Analytics when visitors hit my main consulting service page. I’d like to have a success event called “Adobe Analytics Service Page Views” when visitors hit this page, but as you can see here, I do not:

To do this, you simply create a new calculated metric that has the following definition:

This metric allows you to see the count of Adobe Analytics Service Page Views based upon the Page Name (or you could use URL) that is associated with that event and can then be used in any Adobe Analytics report:

So that is how simple it is to retroactively create a metric in Adobe Analytics. Obviously, this becomes more difficult if the metric you want is based on actions beyond just a page loading, but if you are tracking those actions in other variables (or ClickMap), you can follow the same process to create a calculated metric off of those actions.

Transitioning To A New Success Event

But what if you want to use the new success event going forward, but also want all of the historical data? This can be done as well with the following steps:

The first step would be to set the new success event going forward via manual tagging, a processing rule or via tag management. To do this, assign the new success event in the Admin Console:

The next step is to pick a date in which you will start setting this new success event and then start populating it.  If you want to have it be a clean break, I recommend doing this one day at midnight.

Next, you want to add the new success event to the preceding calculated metric so that you can have both the historical count and the count going forward:

However, this formula will double-count the event for all dates in which the new success event 12 has been set. Therefore, the last step is to apply two date-based segments to each part of the formula. The first date range contains the historical dates before the new success event was set. The second date range contains the dates after the new success event has been set (you can make the end date some date way into the future). Once both of these segments have been created, you can add them to the corresponding part of the formula so it looks like this:

This combined metric will use the page name for the old timeframe and the new success event for the new timeframe. Eventually, if desired, you can transition to using only the success event instead of this calculated metric when you have enough data in the success event alone.

Summary

To wrap up, this post shows a way that you can create metrics for items that you may have missed in your initial implementation and provides a way to fix your original omission and combine the old and the new. As I stated, this functionality isn’t as robust as what you might get from a Heap, SnowPlow or Mixpanel, but it can be a way to help if you need it in a pinch.