Google Analytics has had some data issues over the last few weeks where the statistics for your site were not accurate if you were using advanced segmentation or custom reporting. This was particularly an issue because the data could not be pulled through the Google Analytics dashboard or the API. Google has not stated what caused the issue, but did provide estimated times for repair and also stated that the the collection of the data was never a problem, only the reporting. This caused a delay in client reporting and also now causes a need for deeper analysis in month-to-month changes in the traffic to determine if it truly was a change in traffic or a Google reporting issue. I first planned to hop on my soap box and rant about cloud computing and our vulnerabilities when we use them, but downtime happens for most service providers and this just proves that Google is no exception.
When I noticed the issue with the data, it was due to a message that appeared on the screen when I attempted to run a few custom reports that said “This report is generated in fast -access mode. Learn more.” The click more revealed the definition of fast access mode which I found quite alarming. From the Google site it states”
“In a sampled data report, some report results are estimated from a statistical sample of site visits.
Sampling occurs when you add a secondary dimension to a report and the resulting data for that segment contains more than 500,000 visits. Sampling data allows Google Analytics to generate reports quickly for large data sets.
Data is sampled when you have over a certain threshold of visits in that date range. For example, if you have over 500,000 visits in your selected date range, but were only looking at a single page with around 1,000 views, your visits will still get sampled because we are looking at all your visits data for your site in that date range.”
Sampling data? To provide reports?!? Since the problem with the data has been corrected, this message no longer appears, but I would think that it is safe to say this has not totally gone away and will be back. The problem with sampling data is that it makes all of the conversion metrics inaccurate which is huge if you are measuring ROI or even worse if you are an e-commerce site and attempting to calculate revenue or inventory. Google analytics would not be the only tool that you would use for those functions, but it becomes much more challenging to manage your business effectively when a lot of time is required to analyze the validity of data that you once assumed was accurate.
Also, statistically calculating data could have huge impacts on sites that rely on advertising as their traffic volumes for visits and pageviews could change drastically. Hopefully more details are provided for “Fast Access Mode” before it it fully rolled out and judging by the anger in the Google support forums, there are many others that feel the same way. As I understand this change today, this could cause large swing in traffic reporting that may not even really exist, but could just be calculating incorrectly. Makes me nervous.
If this change does take place, it would also seem that Google Analytics is only a bit better than a Quantcast or Compete.com who also use some type of statistical hocus pocus to estimate site traffic. Really not like the idea of this change…Need more details. Has anyone had problems with your latest Google Analytics data?