Hiii ES community!
I've been stuck on a tricky requirement for quite a while.. So I have a time series data capturing when a user has used what feature as which identity, lie this:
| feature | is_free_trial | trace_id |
| search | true | 000001 |
| register | false | 000001 |
| adv_search | false | 000001 |
The trace_id is used to identify an unique visitor interacting with the web app, and is computed from data like IP and User-Agent.
With this data lying in ES, I am hoping to answer "how many people used free trial features before registering?". I've created a query that answers the question. But I couldn't figure out how to get this metric as a cumulative sum over some time period..
cumulative_sum doesn't seem to work because each bucket interval's min date is fixed, whereas a visitor may trial the application a long while ago (can be months) before registration. So I guess what's really needed is a special "interval" setting that has a fixed start date (e.g., since whatever date the application became live) and a moving end date that grows daily/weekly/monthly?