Call 0845 485 1219
We love digital - Call
0845 485 1219 and say hello - Mon - Fri, 9am - 5pm
As a business owner, it’s likely that you use Google Analytics to track data on your website. Whilst Google Analytics is a fantastically useful tool, there is one area that can cause inaccuracies – the use of sampled data. Find out what this could mean for your website and business.
If you’ve read any of my previous blog posts, you’ll know that I’m a huge fan of Google Analytics; it’s simple, flexible and free! But with all free tools you can expect there to be a few downfalls. As such, we’re going to look into the black hole of sampled data.
Sampled data is when a report in Google Analytics is based on a smaller sub-set of your data as opposed to the full amount of data available. When data is sampled, you are at risk of getting an inaccurate picture of your data dependent on the sub-set of data which is used within the sample. And now breathe. Let’s look at an example to make things easier.
Several metrics within Google Analytics are based upon averages; i.e. average time on page, bounce rate, pages per visit and so on. Everyone knows from their secondary school Math’s experience that the change of one number in a sub-set can have a dramatic effect on an average.
To allow reports to be generated as quickly as possible in its free version, Google Analytics samples data at a limit allowance of 500,000 visits per report (NB: Limits vary dependent on report, see below). Obviously this opens the debate of how accurate the data being provided really is once this upper limit has been met – And we all know how crucial it is that business decisions are made on strong, empirically accurate data.
Sampled data can result in serious business implications. Due to the potential inaccuracy of sampled data, any high level decisions made based on this can result in an outcome which can dramatically effect the performance of your business. For example, using sampled data, Google Analytics shows product A sold 1,200 units and product B sold none. As a result you may be led to the conclusion that product B is no longer worth stocking due to the lack of user interest. In fact product B may actually have sold 1,400 units, but all of these transactions took place in visits which were not included within the sub-set of sampled data. This would result in a huge loss in revenue if product B was cut from the range. This is an extreme example, but as you can see it is important to get an accurate picture before making any crucial business decisions.
In many reports, Google Analytics will show you a notification in the top right hand corner indicating to you not only that the data has been sampled, but also detailing exactly how many pieces of data have been included within the sample. For example,
However, unfortunately Google will not always show you this notification so it is important to be aware of the data allowance limits so you know when your data is likely to be sampled.
• 500,000 visits in most reports
• Flow Visualisation Report – 100,000 visits
• Multi-Channel Funnels Report – 1 million conversion
There are several ways you can address the sampling within Google Analytics to get the most accurate picture of your data. Let’s have a look at these.
In instances of sampling your will be presented with a checkerboard icon in the top right hand corner.
By clicking on this you will have access to a slider rail which allows you to increase/decrease the sample size used within your data sub-set.
To ensure the data analysed within your report is always as accurate as possible, make sure the slider rail is pulled to the Higher Precision extreme. This will take into account as much data as possible within the sample.
When nothing but the full data set is acceptable for your analysis, there are also two other ways you can address the situation. The first can be extremely time consuming, but still involves the use of the free version of Google Analytics. The second is more time-effective, but involves a much higher subscription fee.
For option one, you can simply narrow the date range for which the sampled report shows. By shrinking the date range you can ensure the data shown does not reach the upper limit of the data allowance, therefore, eliminating the sampling from taking place.
Option two, involves the purchase of Google Analytics Premium. This involves a yearly subscription fee and can be an extortionate cost dependent on the size of your business ($150,000 per year). If you are seriously considering this option it is important to take into account what benefits, other than simply unsampled data, this platform can offer you. It is unlikely to be the preferred option for many businesses unless your revenue exceeds $1,000,000 per year.
So, taking into account the importance of basing your crucial business decisions on accurate data, the three options available are to increase the sample size, shrink your date span or pay for Google Analytics Premium.
Copyright © 2006 - 2014, Koozai Ltd