The age old question; How can I get unsampled data from my Google Analytics profile?
Typically if you have a large site on analytics, it can be hard to add filters or segment data without it getting sampled data and, to quote Emma, Senior Insight Analytics Analyst at Edit Agency; “If data is not correct, reports are not correct then decisions are made on incorrect data”, which of course can be a dangerous thing.
What is sampling?
Put simply, sampling is a pain. It means exactly what you think it does; Google only shows you a percentage of your current data which means the rest of the data is filled-in based upon your sample. You still don’t have access to all of your data and the data you have may be inaccurate. Having access to all of your data is essential to paint a picture of how you’re performing on a whole; any partial data will provide you with an inaccurate report and could lead to an inaccurate conclusion.
How do I know my data is being sampled?
It’s easy to tell when your data is getting sampled – just underneath your date range and to the left hand side of your response settings, it will tell you the amount of sessions that your report is based upon.
As you can see (in the above screenshot) it tells you the percentage of your actual sessions that are currently being shown to you, the rest is just automatically filled in based upon your sample, which in this case is 0.67%. What can I do with that? Nothing.
Google samples your data based upon the amount of queries you are sending them, and there are many factors that can effect whether your data is sampled from date ranges to dimensions.
How to avoid sampling:
There are a few basic ways you can try to unsample your data and skate around the issue in order to get your data. We’ve listed a few simple ones here followed by an advanced but more successful technique.
1. Modify your data range
Sometimes the only reason the data will be sampled is because of your date range; the longer or older the date range the more likely it is to be sampled. In order to avoid this, try select a lower date range and export one by one. Again, this can be a pain and take a lot longer than you first intended but it may well get the job done.
2. Replace segments with sources, secondary dimensions, or vice versa
Sometimes Google Analytics can be a bit weird and sample for secondary dimensions and various sources; but not if you apply an advanced segment to do the exact same thing and the same is true for the opposite way round. If, for example, your data for Google Organic Source / Medium is getting sampled, try use an advanced segment instead.
3. Set up new analytics profiles with filters
If you still can’t get any data, try set up a new Analytics profile filtered for the specific thing that you are wanting to get data from. Sometimes this can mean that your data is unsampled but it’s the long way round and may be a bit of a pain if your client doesn’t want to have loads of profiles.
4. Google Analytics Premium, sort of helps….
Google Analytics premium costs a lot of money and it’s only really worth it for a select few large companies. In order to get unsampled data with premium you can go to the export section of the data you require, and get the “unsampled report” option which is brand new.
This is of course useful but it costs thousands of dollars / pounds a year to get the package, and you still can’t view quick snapshots of data within the dashboard as it is still sampled
5. Using the Analytics API
URGH! My data is still sampled?! Don’t lose hope yet! Sometimes when all of these techniques just aren’t enough to cut, and Google is still withholding parts of your data, then have faith as there is still an alternative way. If this is the case then there are also more advanced ways to access the information and it’s through using the Google Analytics API.
The easiest way in which I have found to do this is through using a program called SuperMetrics, which you may be familiar with if you have read any of my previous blog posts, where you can learn how to set it up.
Once you have done this and received the data you want, then you can simply input in the settings section “AVOID_SAMPLING, SAMPLING_NOTE”. This will slow down your spreadsheet significantly however it will provide you with the most accurate data; just make sure that you keep smaller date ranges in order to have a quicker return and reduce the threat of the spreadsheet crashing.