@wikiloops opened this Issue on November 9th 2014

Average generation time is a highly interesting metric in many regards -
p.e. when choosing the right set of server / CDN, or to get an idea how quick the page is perceived by your visitors depending on their location.

When checking average generation time several times a day, I also noticed it seems to vary on certain hours of the day - which in my case indicated a server resource problem caused by some chronjob. I would not have been able to spot it in any other way (again, being on a shared hoster, I am not allowed to look into the server-side monitoring...).

It would be great if there was a report splitting average generation time by country, by browser & device to allow even better evaluation.
It would also be very interesting to compare time-on-website / actions / goals with generation time, I'd assume the later will have an enormous impact on the first.
This is really relevant when deciding on server investments, because it will tell you how much being 100ms slower costs you on the revenue side.

To be able to use it to spot temporary server bottlenecks, introducing smaller date-ranges than the 24hour-minimum we have as also addressed in #6236 would be the thing to do.

@tsteur commented on November 9th 2014 Member


Powered by GitHub Issue Mirror