German, French and English Visits, Internal and External Visits – all in one Report as neat, compact metrics. All this without having to spend a lot of time with tedious breakdowns and segments (and still not getting what you’d really like to). That’s what an Adobe Analytics client needed. The new “Unified Calculated Metrics” now make this possible – something that you can’t do in Google Analytics by the way.
The “breakdown” is something so essential to Adobe Analytics that you can hardly do any real analysis without breaking down. Breaking down is sort of a combination of what Google Analytics users know as Drilldowns and adding a second or third dimension to a report. When you break down with Adobe, you break down one dimension by another one, e.g. the Campaign Name by the Device and that by the Browser and that by the Browser Version and that by the Entry Page and so on… You can get as granular as you want and never have to fear sampling like in GA!
As useful as breakdowns are, there are some limitations to them in the standard “Reports & Analytics” (R&A) interface (the former “SiteCatalyst”):
- Hit-based Dimensions (props) can be broken down endlessly, but only for one value at a time, e.g. only Browser “Internet Explorer” by Version and then only Version “11” by the Entry Page etc.
- For “Almighty Dimensions” (eVars) – they are almighty because they are what you define (they can be Hit-, Visit-, Visitor-, minute-, hour-, month-, until-event-x-happens-based, etc…) – you can break down up to 5 values per page, but only one level deep. For example, you can break down the Campaign Name by Device and see up to five Campaign Names per page with its corresponding Devices, but then you cannot break down one of these Devices by the Browser anymore.
Inside the R&A interface, you can work around this with Hit-based segments (a feature that I dearly miss in Google Analytics!), e.g. you view the report only for a segment where your intermediate breakdown conditions are already fulfilled. To stick with our example, I would use a segment that only includes Internet Explorer 11 Hits from Desktop devices, and then I just view “Campaign Name” broken down by “Entry Page”. It is also quite common to drop most values in a prop AND an eVar so you can have the benefit of both variable types (e.g. endless breakdowns and pathing for props, 255 characters and superpowers for eVars).
If you don’t want to create a segment first for every combination you want to analyze, you can overcome this through the sheer endless flexibility of “Ad hoc Analysis” (formerly Discover), but that amazing tool is just too complex for normal report consumers. The new “Analysis Workspace” (currently in Beta and only for Admins) has added some of the Ad hoc Analysis flexibility to the R&A interface, but it also is a bit too demanding to be used by a regular.
Now, with the totally revamped Calculated Metrics that came out last week, there is a much slicker way to get a lot of these things that you needed to accomplish with the above-mentioned workarounds in the past. And it makes it easy enough for normal report consumers, AND it is faster than breaking down again and again.
The Task: Visits per Site Section per Language
If you are in multi-lingual Switzerland, implementing an Analytics tool usually comes with the additional challenge of three or four different language versions of a site. Even if the visitors are all Swiss, they tend to behave quite differently depending on their preferred language. So the language is one of the most important split parameters for analyses. The client in this example has a website in German, French and English, and he wanted something seemingly easy that turned out to be almost impossible until the new calculated metrics came out (it would be impossible in Google Analytics as well by the way, and it was even impossible with Adobe’s Ad hoc Analysis).
The client wanted a report that showed a dimension (e.g. the Site Section) with metrics for Total Page Views and Visits, with Page Views and Visits per Language, and with Page Views and Visits for Internal/External Visitors – he wanted to see how much traffic is generated by staff (“Internal”) compared to the outside world (“External”).
So the report should look like this (to simplify, I am leaving out the English language here):
Page Views was no problem: For each language, we simply used one of the 1,000 available
Success Events Custom Metrics (something I think of in pain when I have to delete one of my only 20 Custom Metrics in Google Analytics) and count each Page View into one of these Metrics: +1 for the Metric “DE Page Views” if the user views a German page, +1 for “FR Page Views” if the user views a French page etc…
But when it came to Visits, even the most seasoned Adobe Consultants had no solution for this other than setting up a very, very complex form of Event Serialization or taking the data outside of AA into Report Builder, an Excel plugin. Setting the Metric to “Once per Visit” does not solve the challenge either because this setting only counts the Metric the first time during a Visit, independent from the dimension it is counting. So if a user starts on the German homepage and then goes to the German “Women” section, the “once-per Visit” Metric counts only for the Homepage, whereas the “Women” section will show 0 German Visits.
The normal solution for this of course was breaking down the Section by another Dimension for “Language” (which we had). But then you had all the limitations from above. So it would have been a very dissatisfying solution: We could only go one level deep (so we could not see Internal/External in the same report together with the languages), we had only five values per page, it was arguably harder to read, and it took up way more space than if we had had the metrics right next to each other (16 lines for 5 values instead of 5):
In April, when we still had no solution for this problem, Adobe Consultant Jan Exner (see his ugly, but very helpful Adobe Analytics blogs here: English | German) then told us about a feature that he referred to as “Derived Metrics”. It had been introduced at the Adobe Summit in March. He said this might solve our problems, but nobody knew when this feature would be released nor whether it would really solve the problem. So we agreed with the client that we would wait until this feature would come out before we would implement the way more complex and expensive Event Serialization.
Now the feature is here, it is called “Unified Calculated Metrics”, and I would not write this post if it hadn’t solved the problem. It is a feature that we can use to build a lot of reports that look much more compact, are easier to use for the end user and save us and the client from too much segmenting and “breakdowning”.
How does this Solve our Problem?
With the new feature, we can (among many other things) create metrics that only count within a certain segment. So a metric could count Visits (or anything else, e.g. Bounces, Pages/Visit etc.), but only within the Segment “German Pageviews”. (The segment was defined by the dimension “Language” which we set to “de” or “fr” or “en” on each page). The metric is easy to create: Just drag your segment and then the metric onto the Calculated Metrics builder:
Then do this again for each of the other metrics – et voilà: We have new metrics, which – after sharing them with the other users – we can use for ANY report and by simply adding them through the “Add Metrics” button on the right-hand side (not visible in screenshot):
[A careful reader may note that the Visits don’t add up – e.g. DE Visits plus FR Visits is a little bit more than total Visits. This is of course correct and what you want to see, because if a user sees a section page in French first and then changes the language to German (the site may have misidentified him as a French speaker first), then the Visit will count for both French and German. In total it will still be just one Visit – a classic example of “never add up Visits”.]
Summary: Simple, but complex
The client request seemed so simple and basic in today’s world where we talk about much more “advanced” concepts and requirements every day. But often, it is the simple and basic things which are so tedious that you can never get to these “higher spheres”. And it showed me again how powerful Adobe Analytics can be. To say it again, the Unified Calculated Metrics, a feature unique to AA, would not be so powerful if Adobe didn’t have another very powerful and unique feature – the unsampled hit-based segments (Google only offers Visit- and Visitor-level segments, and those only with lots of sampling). In Webtrends Explore, by the way, you could do the same thing on the fly, but Explore is currently a bit like Adobe Ad hoc Analysis – completely separate from the rest of the reporting, and a bit too advanced for normal report consumers.
Share your Experience
Have you experimented with the new Unified Calculated Metrics already? Do you have other favorite features in Adobe Analytics that you don’t get in Google Analytics or other tools?