7 Steps to Clean up Adobe Analytics in a Data-Driven Way
Merge duplicate segments and calculated metrics, delete unused components, and get rid of those thousands of old workspaces
Thanks to the Adobe Analytics API, you can finally get your Analytics setup back under control and put an end to user (and admin) frustration because of seemingly endless lists of components with similar names, duplicates etc.
A good data engineer can write the queries needed for the steps below herself with the Adobe Analytics API 2.0, e.g. via Julien Piccini’s Python Wrapper. Or you can go the easy way and use the Adobe Analytics Bulk Component Manager for Google Sheets by Datacroft (see previous articles on the Component Manager as a driver for efficient Virtual Report Suite Components Curation, or how you can easily find 100’s of unused segments with it).
Step 1: Identify unused Workspaces
This is the only part of the process where you need a little help from your end users. Get a list of all workspaces that have not been modified in the last 6 months. In the Component Manager, you would go to the “Workspaces” tab. Update it via the menu “Adobe Component Manager > Workspaces > Update Workspaces list”, then filter the “modified” column by “Date is before” e.g. “2020–07–01”):
UPDATE: You can now also additionally filter by “activeSchedules” = 0 to leave those Workspaces that still have active scheduled emails. As these schedules have to be actively renewed after a year, chances are high that workspaces with an active schedule are still needed.
Step 2: Tell owners to mark the Workspaces they still need
Copy the filtered list to another GoogleSheet or download as an Excel file, then create a new email with “ACTION REQUIRED UNTIL [[your deadline]]” in the subject line. Do a UNIQUE(“owner email” column, e.g. B3:B) in a free column to get all distinct email addresses. Copy them into your email recipient field.
In the email to the owners, send the link to the sheet, explain the benefits (cleaner, faster Analytics with less interpretation errors, misunderstandings and frustration) and explain that they need to mark all Workspaces they still need with “x”.
Step 3: Delete all unneeded Workspaces with one command
Now, with some simple GoogleSheet/Excel formulas (I trust you here), you invert the users’ x’s (x becomes empty, empty becomes x). Now, x marks all the Workspaces to delete. Then run “Workspaces > Delete Workspaces” and get a glass of kvass. Done!
When the deletion is done, the “Workspaces” tab refills, with now hopefully hundreds of Workspaces less (at 2 clients, we recently deleted 700 and 1'230 Workspaces respectively this way). Moreover, in the “update_log” tab, you can download an Excel file that shows you which workspaces were deleted (un)successfully (unsuccessfully e.g. because the Workspaces had been deleted before already).
Step 4: Identify unused components
Now that we have cleaned up the Workspaces, we can update the “all_comp_usage” tab by running “Component Usage > Update Component Usage Tab”.
The Component Usage tab is probably the most useful of all (see this video to see it in action). It shows all components by their usage, i.e.: In how many and in which Workspaces, Segments, and Calculated Metrics is each component used? See e.g. the Calculated Metric “Avg. Order Value (Product Revenue)” marked in green below: It is quite popular, as it is used in 146 Projects (Workspaces), 0 Segments and 0 Calculated Metrics (which is no surprise because Calculated Metrics cannot be used in Segments or other Calculated Metrics).
This is super useful, because we can now simply filter like this:
- componentType “calculatedMetric” or “segment”
- includeType: “all” or “shared” (because “template” components refer to out-of-the-box stuff like “Visits” or “PageName” which cannot be changed)
- matches_total: 0
This gives us a list of only the components that are not used anywhere and can likely be deleted. This list should now be longer — because we have just deleted a lot of workspaces where a certain component may have been used before.
Step 5: Delete all unused Components with one click
Now copy the IDs, switch to the “component_editor” tab and paste them anywhere into the “id” column in the orange “EDIT” area. The other columns will fill automatically.
Now choose method “delete” in one of the drop-down menus in the “method” column and copy it down to the last row.
Now, choose “Component Editor -> Send Updates to AA”.
You are taken to the “update_export” tab and can review the changes before finally confirming.
Need I mention that the update_log tab logs all these deletions as well? No, I don’t. The status area on the component_editor tab shows you the progress and reloads all components after the deletion is finished. You can now verify in the left blue area that the deleted components are indeed gone (Select “full_comp” for All Components and then search for the deleted IDs — or go into Adobe Analytics and see for yourself! :))
Step 6: Identify duplicates
The Component Usage Tab also shows you the duplicates for all segments and calculated metrics! Often, users e.g. create a new segment even though a segment with the same definition already exists — that of course happens more often in badly curated setups without proper Component Curation in Virtual Report Suites. A consultant recently told me that one client has amassed so many segments that the Segments list in Analytics takes so long to load that they simply create a new segment every time (because finding out what is already there is so tedious).
Now you can help your users by automatically replacing duplicate components by their “default” version:
- Identify the duplicates by filtering out blanks from the column “duplicate_calcmetrics_names”
- Sort the table by the same column so you have the duplicates on top of each other.
- Determine the “survivor” for each duplicate.
Example: In the first 2 rows below, we see 146 usages for “Avg. Order Value”, but only 3 usages for its duplicate “Product Revenue / Orders”. Here it is clear that “Avg. Order Value” deserves to survive the cleanup. You can use the “remarks” column on the left (not in screenshot) to mark the survivors with S and the losers with L.
Step 7: Merge duplicates by automatically replacing them everywhere
Now go to the latest feature of the Component Manager, the Component Replacer. It goes through all Workspaces, Segments, and Calculated Metrics and replaces a component by another one, e.g. eVar28 by eVar30 or Segment X by Y.
Copy-paste the IDs of the “losers” (the components that will perish) on the left side and the “survivors” on the right side.
In the example below, “% Bot Traffic” will e.g. be replaced by “% Visits by Bots”. Now run Component Usage > Run Component Replacer and wait a bit.
After the cleanup, we can delete the components on the left (the losers). Once again, we simply copy their IDs over to the “component_editor” tab and follow “Step 5: Delete all unused Components with one click” to delete them.
(optional) Last Step: Now tackle your technical implementation (eVars, props, Success Events)
Now apply the whole thing to your technical implementation (Adobe Launch, Report Suite configuration): Just like for segments, the Component Usage tab shows you whether your highly regarded eVar127 is used in any Workspace, Segment or Calculated Metric (apart from maybe your own ones). Especially Success Events (Custom Metrics) have a tendency to be plentiful in size, but scarce in actual usage. Here we see e.g. that Event 121 and 119 are not used anywhere, whereas Event 40 is only used in some QA Workspaces. So we can probably scratch them from our Tag Management System, Processing Rule and Report Suite Configuration and make your whole implementation leaner, and that of course means less causes for errors and easier to test and monitor.
Tedious manual sifting-through is in the past
That’s it. So much more lightness and order for your users in your Adobe Analytics. No more manual sifting through thousands of segments, metrics or Workspaces, wasting weeks with it and never getting done. And I have not even covered all of the Component Manager for GoogleSheets by Datacroft features yet! You can e.g. also use it to efficiently add/remove/rename curated components for many Virtual Report Suites at once, another must-have if you want to take data democratization seriously. So it is not only a tool for a big cleanup, it is something I use on a daily basis.