Including rows outside of dimension filter for running total

[Maikerusan]

I’m currently tasked with getting a cumulative running total of the number of users that log in to my companies site for metric purposes. The Dashboard that exists simply pulls in all the user login event data (this includes a variety of information such as the client and other data but the only relevant data is the login date) and uses the running_total method along with some grouping to display the log in data.

There is also a filter on the login date dimension that can accept a start date and end date as parameters to limit the scope of the data being viewed instead of across all time. I know from the documentation that filtering on a dimension reduces raw data before calculations but I was wondering if there was a proper Looker way to include data that is being filtered out as part of the cumulative total to display in the visualization.

An example of the problem I’m running into is like so:
If I use dates that cover the entirety of the data set and I have 5 user logins on Day 1 and 5 user logins on Day 2 and 5 user logins on Day 3, the cumulative logins at Day 1 would be, at Day 2 would be 10, at Day 3 would be 15 and so forth.
When I filter to have a start date that starts on something like Day 2, the cumulative at Day 2 would be 5 and at Day 3 would be 10 (missing the first 5 logins from Day 1 because the filter cut it out).

Ultimately the best way I would be able to display this data would be for the raw data to have the cumulative logins across all time and the visualization to only show the data for the date range in the parameters that are passed in as the filter.

[Guilherme_Ferna]

I am having the same issue.
Hope this helps: How to filter a measure but still show totals

[Maikerusan]

I saw this I was researching the problem initially. It’s close to what I need but the problem I ultimately run into then is there would have to be a filter on a calculation column or something like that where the people who are creating the reports would then need to know the running cumulative to filter on, at which point the date would become a pointless dimension. What I’m doing right now is recreating the look with all custom filters to make sure the rows still show in the results, then I use a yesno column for each dimension and create a final yesno that I use for calculations which is a combination of all the prior yesno columns. It gives me the solution I want but doesn’t feel great 😕

[Guilherme_Ferna]

I’m finding Looker a little limited. But, to be honest, I don’t have enough experience yet: D