Analytics
4.1K members online now
Understand information in your reports and troubleshoot reporting issues such as self-referrals, (not set) data, and inaccurate information
 
Guide Me
star_border
Reply

Anomalous pageview data - multiples of 16

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

We've encountered a very strange data anomaly in our GA data and so far I haven't found any suggestion as to what might be causing this. When pulling page-level data for a 13-month window, we are seeing low numbers appear in multiples of 16. There are literally no pageview numbers below 16. I have a screenshot of the data table showing some of this data. This occurs when drilling down into anything far enough that the numbers drop below 300. At that level, every figure and below is a multiple of 16, which doesn't make any sense at all.

 

When changing the date range to a 12-month window, instead of finding multiples of 16, I find multiples of 15 in numbers 15-60, and at figures above that and below 300, they are multiples of 15, plus one, i.e. 61, 76, 91, 106, 121 etc. There's a screenshot of this below as well.

 

I'm stumped - it appears to be a math error within Google Analytics, but what could be causing it? Implementation problem? CMS or hosting issues? Really need a Google rep to help us!

 

This is looking at a 13-month window, April 1, 2016 - April 30, 2017.This is looking at a 13-month window, April 1, 2016 - April 30, 2017.This is looking at a 12 month window, April 30, 2016 - April 30, 2017.This is looking at a 12 month window, April 30, 2016 - April 30, 2017.

 

 

1 Expert replyverified_user

Anomalous pageview data - multiples of 16

Rising Star
# 2
Rising Star

I'm pretty confident this is a sampling issue, particularly given the presence of a secondary dimension in your data. Check to see whether you are viewing 100% of your data, or a sampled sub-set - this information appears by the date selector. 

John Wedderburn, Advertiser Community, Rising Star
Was my response helpful? If yes, please mark it as the ‘Best Answer.’ Learn how here.

Anomalous pageview data - multiples of 16

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭

I think you're exactly correct John. Thank you for your response - I hadn't considered that GA would use a sample data set small enough that the data would be so unreliable. But now I see that the sample is very small with a secondary dimension applied. Is there a way to force GA to use a larger sample for the sake of accuracy and usable figures?

 

Also, since the small numbers are full of errors because of this sampling error, does it mean also that the larger numbers can be trusted? I'm concerned that even my numbers above the error threshold (200 or 300) aren't reliable.

 

Thanks!

Marked as Best Answer.
Solution
Accepted by topic author Novant S
May

Anomalous pageview data - multiples of 16

Rising Star
# 4
Rising Star

Hi - good to know we're getting towards a solution. Sampling is an issue in GA - and definitely worth understanding. There's plenty of resources out there, but these contain some good info:

http://www.blastam.com/blog/can-you-trust-your-google-analytics-data ( a bit old, but there comparison data is interesting)

http://www.lunametrics.com/blog/2016/03/03/accurate-sampling-google-analytics/ (similar to above, but more in depth with confidence interval calculators)

https://support.google.com/analytics/answer/2637192?hl=en (overview from GA)

http://www.lunametrics.com/blog/2013/06/24/solutions-google-analytics-sampling-problems-8-ways-data/ (how to solve this issue).

 

Personally, my experience has been - as suggested in the links above - that your coarse grained stats, such as sessions and pageviews, can be trusted but events such as transactions can be very different from reality; especially when you're only looking at 50% or less of the data. To answer your questions, I'd be reasonably happy with using the larger numbers but the smaller, duplicated ones are sketchy. 

 

If I were you, I'd get a handle on the impact of the sampling by comparing unsampled with sampled data. The easiest way to get unsampled data is to chop your date range into small enough chunks so that sampling is no longer an issue - this can be laborious so the API can make this a little easier (use the API add-on in Google sheets). Lunametrics describe this method  as #8 in the article I linked to.

 

You can force GA to use more data, by using the greater precision selector but you'll probably find that this has a limited effect.

 

 

 

 

 

John Wedderburn, Advertiser Community, Rising Star
Was my response helpful? If yes, please mark it as the ‘Best Answer.’ Learn how here.

Anomalous pageview data - multiples of 16

Visitor ✭ ✭ ✭
# 5
Visitor ✭ ✭ ✭

John you are awesome! Thank you so much for helping me to understand what's going on here. Based on the links you provided and your assessment, I not only know exactly why I'm seeing this anomalous data, but also have a solution to get the info I need. All the best to you!