It takes a few minutes for the services to warm up the statistics cache after update or a new installation. 
However, for newly added things, like a new dataset, information appears in a few seconds. There are still some factors that may mislead you if you are interested in “real-time” data.

  1. When a graph is rendered, it’s horizontal time-scale is “now” minus 1 minute (for example). “now” = time in the browser / client machine
  2. When a graph is requested from the analytics service, it’s horizontal data points are for “now” minus 1 minute (for example). “now” = time on the storage

Even if your browser and storage are synced to the second, there will be a 2-3 seconds error/delay because of the time needed for the request to round-trip to the storage.

Because of the problems above, there can be two different effects on the graphics:

  1. If browser time is ahead of storage time, graphics will be shifted to the left. For example if you are looking at the last 5 minutes, and your computer is 2 minutes ahead of the storage, and the current time is 00:05, you will see the graph rendered for 00:00-00:05, and the graphics for 00:00-00:03. Because the storage is 2 minutes behind you, it will provide data for 23:58-00:03, but you render only what is between 00:00-00:05.
  2. If browser time is behind of the storage time, graphics will appear with delay. For example if you are looking at the last 5 minutes, and your computer is 2 minutes behind of the storage, and the current time is 00:05, you will see the graph rendered for 00:00-00:05, and the graphics for 00:02-00:07. Because the time on the storage is in the future, compared to the browser, the data on the graphics will be 2 minutes late from your point of view.

If the date and time is synchronized, you see data with a very good precision and only 2-3 seconds delay at most.