Suppose I have the following breakdown imported from Data Warehouse
Dimension 1 | Dimension 2 | Recipe Id (Target) | Page Views | eVar Instances |
---|---|---|---|---|
A | D | B | XXXX | XXXX |
A | E | B | YYYY | YYYY |
A | F | G | ZZZZ | ZZZZ |
then I decide to remove Dimension 2 because It's not relevant for what I want to calculate, so I remove the column and aggregate the rows (using a tool like Excel or Python), leaving me with the following table
Before aggregation:
Dimension 1 | Recipe Id (Target) | Page Views | eVar Instances |
---|---|---|---|
A | B | XXXX | XXXX |
A | B | YYYY | YYYY |
A | G | ZZZZ | ZZZZ |
After aggregation:
Dimension 1 | Recipe Id (Target) | Page Views | eVar Instances |
---|---|---|---|
A | B | XXXX + YYYY | XXXX + YYYY |
A | G | ZZZZ | ZZZZ |
is this a valid operation? My intuition tells me that for Page Views and Instances this is valid since the rows are exclusive. On the other hand, if I had a Visits metric it may be invalid since XXXX and YYYY may share a same visit.
How about Recipe Id (Target) or any other Target dimensions? A hit may belong to many Target Activities (AB tests for example), but Target breakdowns are individual, so if I want all Dimension 1 Page Views for element A, it may be wrong to sum XXXX + YYYY + ZZZZ because there is a chance that all Page Views of A belong to both B and G, resulting in inflated numbers of Page Views.
Thanks for any help, I'm still learning about these concepts
Solved! Go to Solution.
It depends on what metrics you're using. If you're using anything that is not hit-based, like visits or unique visitors, they would not add up. If you're using hit-based metrics like custom events, they'll likely end up pretty close.
It depends on what metrics you're using. If you're using anything that is not hit-based, like visits or unique visitors, they would not add up. If you're using hit-based metrics like custom events, they'll likely end up pretty close.