Google Analytics – Faster Processing or Higher Precision – is it a feature?
Google Analytics is introducing a lot of smaller UI changes and new features lately. One of them only shows up when looking at graphs of the current day. You can chose between fast processing and high precision.
And then there’s caching..
Figures of the current day or even the past two days have never been very accurate. This has to do with the way Google Analytics deals with these vast amounts of data via map/reduce. There simply is no real time access to the raw data. Instead data is mapped and reduced to fill many smaller data pools which can be used for the reports.
So when looking at the current date the mapping and reducing is simply not done yet. For the current day it can not be completed and for the past day I assume that the results are not distributed to all servers.
So for the current day Google came up with this choice. Either you get a snapshot that might be an estimate from only a small part of your data but you get it fast. Or you can get a very accurate number that is close to real time but it will take a fair bit of processing.
Sounds reasonable right.
Well it would be if it would work this way.
It might be caused by my blog not being a traffic rich website but rather small. But I got the impression over the past few days that the numbers resulting from either choice are cached rigorously.
In my experience the figures increase whenever I change this control no matter in which direction.
Even if I choose a higher precision and a few minutes later do a reload the figures will not change. However if I then change to faster processing the figures increase. So I conclude that the high precision data comes from some kind of cache.
This experiment works the other way around to. Figures seem to get cached for a certain time during which you can only get more current figures by switching the control to something else.