You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Firebase is reputed to discard events that are older than 72 hours. The original source of this information appears to be a Google presentation in 2019. Apparently older Crashlytics reports have their timestamps rewritten on the client side to sneak them past the server's 72-hour limit!
We believe that there will be a number of Endless Key users who use the app offline for more than 3 days, but are online weekly or so. A way to measure this might be:
Have the app monitor its connectivity state
When it first goes offline, record the timestamp S
When it next notices it's online, at timestamp T, if T - S is greater than 72 hours, record a custom event (say) offline_for_more_than_3_days with value round((T - S) / 24) or similar
Then we can count the number of users associated with this event in a given time period, divide it by the number of unique users in the time period, to determine what % of users are in this boat; and we can check the distribution of time between using EK online. (This is not the same as the frequency at which the user is online because the app may not be running!)
The text was updated successfully, but these errors were encountered:
Firebase is reputed to discard events that are older than 72 hours. The original source of this information appears to be a Google presentation in 2019. Apparently older Crashlytics reports have their timestamps rewritten on the client side to sneak them past the server's 72-hour limit!
We believe that there will be a number of Endless Key users who use the app offline for more than 3 days, but are online weekly or so. A way to measure this might be:
offline_for_more_than_3_days
with value round((T - S) / 24) or similarThen we can count the number of users associated with this event in a given time period, divide it by the number of unique users in the time period, to determine what % of users are in this boat; and we can check the distribution of time between using EK online. (This is not the same as the frequency at which the user is online because the app may not be running!)
The text was updated successfully, but these errors were encountered: