Analytics
5.5K 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

Segments & "All Users" Inconsistencies

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

Hello everyone, 

 

In reporting > Audience > Overview when I have the month of August selected and "All Users" there are X sessions. When I include US and Ex-US segments the number of sessions for "All Users" changes to 3X This doesn't happen any other month. Additionally if I look at the number of users on 8/14 and I have a date range of 8/1-8/20 selected it is Y but if I have 8/1-8/31 selected the number of users on 8/14 is 2Y. 

 

In terms of changes to the website, there was a data import on 8/17 and there were new tags published on 8/25. Also note that we upgraded our analytics tracking code to include some custom tracking parameters

 

I have no idea what could have caused this and need help troubleshooting. 

 

Thank you! 

 

1 Expert replyverified_user

Re: Segments & "All Users" Inconsistencies

Rising Star
# 2
Rising Star
Hi Caroline,

It sounds like you are getting sampling when applying the segments -- which would be expected with that volume. You probably have some days in August that have spikes of traffic and these sessions are getting included in the sample and extrapolated over the entire period. I don't know what else is in your segment but if it's a simple geo you could use the location reports to get clean data.

Best,

Theo Bennett
Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Re: Segments & "All Users" Inconsistencies

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
Hi Theo!

Thanks for your help with this. After some further exploring I can see how this could be caused by sampling in some cases but I don't believe that is true in this instance. We have been collecting this data in the same fashion for a few years and haven't seen this before. I have provided several more specifics below.

1. Reporting -> Audience -> Overview
- "All Users" Segment - Okay, on trend with past months and no spikes in data throughout the month
- "All Users" Segment & "US" segment (which is country matches exactly "United States") -> All users data increases by a factor of 2.75. Adding this segment over any time period pre mid august doesn't change the all users data.
- "All Users" & "Age >45" -> all users increases by a factor of 2.75. Adding this segment over any time period pre mid august doesn't change the all users data.

2. Reporting -> Audience -> Geo -> Location
- "All Users" segment -> okay
- "All Users" segment & "Medium" secondary dimension -> all users increases by a factor of 2.75.

3. Custom reporting
- "All Users" segment & dimension drilldown = "Date" -> okay
- "All Users" segment & dimension drilldown = "country" -> all users increases by a factor of 2.75.

In summary, as soon as I start to try and customize my reports, whether in reporting or customization, the values change. Customizing unfortunately is very important for how we analyze our data.

I am mainly confused as to what changes we could have made that would effect All Users data ONLY when using segments. We added user ID tags, change is state tags and timer tags. All were set to non-interaction. The code that we added is below:

<script type="text/javascript">
(function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){
(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o),
m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)
})(window,document,'script','https://www.google-analytics.com/analytics.js','__gaTracker');
var uuid= '<?php echo $_COOKIE["mdc_uuid"]; ?>';
__gaTracker('create', 'UA-XXX, 'auto', {'userId': '<?php echo $_COOKIE["mdc_uuid"]; ?>',}); // Set the user ID using signed-in user_id.
__gaTracker('set', 'forceSSL', true);
__gaTracker('require', 'displayfeatures');
__gaTracker(function(tracker) {var clientId = tracker.get('clientId');});// get client ID
__gaTracker('send', 'pageview');

&lt;/script&gt;


If there is any other information that we can provide to help diagnose this issue please let me know.

Re: Segments & "All Users" Inconsistencies

Rising Star
# 4
Rising Star

 

Hi Caroline,

If you are not on GA Premium/360 you should get sampling when you apply that segment or add in a secondary dimension and in your custom reports... basically anytime that you are asking GA for something that is nor pre-aggregated. See the screenshot below; anything less than 100% means you have sampling.

 

ss sampling.png

 

Best,

Theo Bennett

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Segments & "All Users" Inconsistencies

Visitor ✭ ✭ ✭
# 5
Visitor ✭ ✭ ✭

Hi Theo,

Thanks so much for your help with this. You are right, after looking into it a little bit more it was a sampling issue. We were sampling at about .5% compared to the 50%+ that we were sampling for previous months. We removed the code and tags that we added and everything went back to normal. Then we added the code back and everything is still fine. This means it must be a tag. Do you know of anything that we could be collecting that would effect sampling? The number of sessions is very similar to previous months and therefore it must be something else. What effects sampling other than the number of sessions? We would like to add these tags back in but definitely need to figure out why they're cause the sampling to drop by so much.

Segments & "All Users" Inconsistencies

Rising Star
# 6
Rising Star

Hi Caroline,

 

There are two types of sampling:  at the collection level and at reporting.  I'm not referring to collection - which is tag based.

 

In reporting, you have pre-aggregated tables that Google Analytics (GA) constructs for you.  Anytime you ask GA for something different - like adding a secondary dimension, apply segments (which you did in your original post) or use a custom report; and there are more than 500,000 sessions in your date range -- GA will not return all the data but will sample some of the users over the date range and extrapolate the results.  That's why you see different results, it most likely has nothing to do with your tagging.

 

More on sampling.

 

Best,

 

Theo Bennett

 

Analytics Evangelist at MoreVisibility | Contact Me
Connect on LinkedIn

Segments & "All Users" Inconsistencies

Visitor ✭ ✭ ✭
# 7
Visitor ✭ ✭ ✭

The issue that we ran into after applying tags was that our sample dropped from > 50% down to 0.5%. Once we removed the tags that we had applied the rate jumped back up to > 50%. Therefore something we did caused the sampling (on the reporting end) to drop even though we did not see an increase in sessions. So, when GA goes to choose how much they will sample is it strictly based on sessions or are there other factors? If we can pinpoint what other factors contribute to sample size we can make sure to not adjust them and then maintain a sample size of around 50%. Additionally when I look at the data for a single day (when we have <500,000 sessions) we still see sampling at around 1% BUT only for the time in which we had applied the tags.