Analytics
1.8K 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

Events report: Discrepancy between sampled and unsampled report

[ Edited ]
Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

Hi, all

 

I'm experiencing an issue when viewing a sampled report of a event that I'm tracking, and I would like to know if it's caused by the sampling.

 

I'm generating a report to view the number of outbound link clicks for a specific domain. When I view the report for February 2016, the total number of events is 15, shown both on the chart and in the table. However, when I increase the date range to February - April 2016 (at which point sampling is applied), the total number of events in February increases by two points to 17.

 

Has the increase occurred because the total event number is estimated in the sampled report?

 

Thanks

Steven

1 Expert replyverified_user

Re: Events report: Discrepancy between sampled and unsampled report

A T
Participant ✭ ✭ ✭
# 2
A T
Participant ✭ ✭ ✭
if the report is sampled it will show a yellow notification in the report in GA.

For these small numbers it doesn't seem sampling issue but more of way the calculation is done from GA

Re: Events report: Discrepancy between sampled and unsampled report

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
Hi, thank you for your response.

I can see when a sample is applied.

What I would like to know is whether or not samples are used to estimate the total metric number?

Thanks
Steven

Re: Events report: Discrepancy between sampled and unsampled report

Rising Star
# 4
Rising Star
Steven to answer your question yes if sampling is taking place the actual counts are then based on the subset of data that was used to get the sampled counts, Sampling is generally based on the number of sessions within the reporting period (date range selected) this sampling occurs at the Property level and not the view level. A second type of sampling occurs when you ask for an ad-hoc report during this sampling it occurs at the property level first and then at view level, so when you asked to increase your date range you asked for an ad-hoc report that is filtered in which case the sampling algortyhm kicked in and sampling occured.

If you want to get an accurate count do each month between February and April by itself and you will most likely not incure sampling, you can then add up the individual months and get a much more accurate count of your data.

Re: Events report: Discrepancy between sampled and unsampled report

Visitor ✭ ✭ ✭
# 5
Visitor ✭ ✭ ✭

Hi Brian

 

Thank you for your response. So to confirm, when sampling occurs, the total count = the estimated total count. Or is the total count of a sampled report the total count within the sample?

 

Because the total count for February increased by two points in the sampled report, my assumption is that it was the estimated total based on the sample.

 

I think I'm getting it!

 

Regards
Steven

Re: Events report: Discrepancy between sampled and unsampled report

Rising Star
# 6
Rising Star
It is an estimate based upon the sampled data, I think you are getting it.