Analytics
1.7K 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

Organic traffic discrepancies when I add a second dimension

Visitor ✭ ✭ ✭
# 1
Visitor ✭ ✭ ✭

So...I've got a bit of a problem.

 

When I view my client's data through the source/medium filter, and then set it to show just organic traffic, it's shows one number for visits (15,711). When I add "Landing Page" as the second dimension, it shows a completely different number (15,655). Not too far off. But when I compare to the previous period, for just organic, it shows May had 15,711 visits and April had 15,735 (up .23%). K, now I add the secondary dimension and the numbers change; May had 16,989 and April had 13,443 (up 26.38%). 

 

Now if I compare April to March, it shows that April had 15,168 with the second dimension applied and 15,279 with just the organic dimension showing.

 

The numbers are all over the place! Has anyone else seen this happen? What could be causing it?

Re: Organic traffic discrepancies when I add a second dimension

[ Edited ]
Follower ✭ ✭ ✭
# 2
Follower ✭ ✭ ✭

Hi Tyson,

 

the discrepancy you are seeing is due to a sampling error in the metrics Google Analytics is showing you.

 

In fact, Google Analytics sometimes computes the report metrics based on a sample of sessions, i.e. a subset of all the sessions.

 

Sampling usually is automatically applied by Google Analytics when you add a secondary dimension / filters / segments, and the overall session counting for the selected period exceeds a certain threshold.

 

When sampling is applied, you will see on the top right part of your screen a square icon with smaller squares inside, like the following screenshot:

 

Schermata 06-2457557 alle 18.44.28.png

 

If you click on it, a switch appears and you can then increase or decrease the sample size, for increasing / decreasing the accuracy in the metrics shown in the report.

 

 <edited by a Community Manager as noted below>

 

 

You can find more info here:

https://support.google.com/analytics/answer/1042498?hl=en&ref_topic=2601030

 

Hope it helps!

 

Best,

Gabriele

 

Note: this post has been edited by a Community Manager, per Community Posting Guidelines. This was done in order to remove a screenshot that included private information such as an email address, an account ID, or similar 
 
Please feel free to re-post the screenshot after having removed or entirely obscured any and all information that has the potential to put your accounts and/or your privacy at risk, when published in a public forum. Thank you for your understanding.

Re: Organic traffic discrepancies when I add a second dimension

Visitor ✭ ✭ ✭
# 3
Visitor ✭ ✭ ✭
This does help! Thank you!

So would switching to Analytics 360 provide more accurate data? If we don't switch, which are the more accurate numbers to go off of? Just "Organic" or the secondary dimension?

Re: Organic traffic discrepancies when I add a second dimension

Follower ✭ ✭ ✭
# 4
Follower ✭ ✭ ✭
Hi Tyson,

you're welcome!

Switching to Analytics 360 would give you a far larger threshold of sessions before seeing sampled data in the GA reports, so, unless your client receives a huge traffic on his site (like millions of sessions per month) it is unlikely that you will generate sampled reports; moreover, you would still be able to export unsampled data in CSV format (or other likewise formats), regardless of the number of sessions in the selected period.

If you don't switch, the simple reports (i.e. no filters, no secondary dimensions added) are usually unsampled. So you should see traffic sources report always without sampling.

However, when you want to see two dimensions in the same report (e.g. traffic sources as primary dimension and landing page as a secondary dimension), you will probably see a sampled report.

Honestly, I would not worry too much of sampling. When the report is sampled, the Google Analytics sample size is fairly large (500k sessions having moved the switch on the top-right towards the maximum accuracy), so the error in the metrics should always be small enough to take meaningful decisions.

Hope it helps!

Best,
Gabriele