Sampling in Google Analytics helps speed up the response from GA, but it can serious affect the accuracy of the data you are looking at.
Fortunately, Google Analytics has a great API, and with Analytics Canvas you can use a visual interface to get at your data, know when sampling is happening, and in many cases eliminate it. In our latest version, we’ve added what we think will be a handy visual warning if sampling is detected.
Detecting sampling- look for the Yellow
It was always possible to detect sampling in Analytics Canvas, but now, when you make a query block, it will automatically check to see if sampling has occurred, and will warn you with a yellow exclamation mark right on the block. Don’t let sampling sneak up on you!
If you see the yellow exclamation mark, you know that the number of visits within the query exceeds the limit (which is generally 500,000 visits) and that sampling has been reported by the API.
In some cases, sampling does not effect your results too much- in others, particularly when segmenting and wanting to look at the long tail data, it can make the results returned very inaccurate.
Analytics canvas lets you eliminate sampling by automatically breaking the query into multiple smaller queries that are not sampled, then automatically combining the data back together so you get the result you wanted. You can do this by simply specifying the maximum number of days you want to query at a time- then set any date range you want, and Analytics Canvas will do the rest.
You can learn more about exactly how this is done here. So if you’ve got buckets and buckets of data and want exact, not sampled data, then give analytics canvas a try.