Misunderstood Metrics: Google Analytics Users

Why do the Users numbers in my report not add up to the total in Google AnalyticsThis Misunderstood Metrics article explores the Users metric in Google Analytics, why the numbers are different, and why both of them are probably wrong.

Adding Up Users: Don’t

TL;DR  You cannot simply add up Users numbers from various days (or countries, or sources, or…) and expect the total to match a total user count for your site. Simple example (see image below), 457 users visited on the 12th and 392 visited on the 13th. The sum is 849, but the report shows the Users count for the 2 days is only 816. That is because 33 people visited ON BOTH DAYS, and a sum is counting them twice (816 actual 2-day total + 33 users visiting both days =  849 summed). NEVER add Users numbers together to get a total.

Measuring Users in Google Analytics

I know, simple example, but there’s more to consider…let’s look at what defines a User in Google Analytics. In most cases, it is a visit from a browser with a cookie containing a unique user id number. If there was no cookie, then a new user id (and a new cookie) would be generated, and it would be considered a visit  from a ‘new’ user. If there was a return visit from the same browser with the cookie, Google Analytics digs up the user id from the cookie and links that visit to the same ‘User’.

Bad Assumption: ‘New Users’ are New Users

With that definition, the real-life User number is flawed by design — if a person clears their browser cookies, uses incognito mode in their browser, or if they use a different browser or different device (like a tablet),  they will appear as a ‘new’ and unique user. In advanced implementations of Google Analytics, you can reduce that problem by implementing a cross-device User-ID in your website. But without that, some of your ‘new’ users could actually be returning users, and so the total Users count in any report is probably over-inflated a bit.

Multiple Sessions Per User

While it may seem obvious, a single user can have multiple sessions represented in any report you look at. They could enter your site from a search, then come back an hour later direct, then the next day from an email campaign. They could be travelling and appear on Sunday from Brussels and on Monday from London.

This is why you can NEVER add up the Users numbers in your report to get a total. Individual users could be counted in multiple rows and your total would include duplicates. if you want a total, get a total unique user count, not a sum. A total unique user count can be obtained by asking for ‘Users’…that’s it, no Date column, no Source, no Country…just ‘Users’.

Getting Users Numbers That Match

So when you are building you own custom report, how do you get Users numbers that match Google Analytics? Like this:

Audience > Overview Report

 

Users Chart Numbers
Dimensions: Date (or Week, or Month)
Metrics: Users
Date range: same as report

Users Widget Number
Dimensions: none
Metrics: Users
Date range: same as report

Audience > Geo > Language Report

 

Table of Languages
Dimensions: Language
Metrics: Users (and other metrics)
Date range: same as report

Total at top of column
Dimensions: none
Metrics: Users
Date range: same as report

Acquisition > All Traffic > Channels

 

Table of Channels
Dimensions: Default Channel Grouping
Metrics: Users (and other metrics)
Date range: same as report

Total at top of column
Dimensions: none
Metrics: Users
Date range: same as report

I think you get the idea — match the dimensions EXACTLY as you see them in the report. If there is a total, it is NOT a sum; it is a separate query made without any dimensions.

Summary

If you want the Users number in your report to match the numbers you see in Google Analytics, match the dimensions exactly. Never add up Users numbers to get a total — make a separate query without any dimensions.

And don’t get too hung-up on how accurate the number is — by design, it is probably wrong (too high) anyways, and no one wants to suggest that there are actually fewer users on your site than the report suggests…

Discover Other Misunderstood Metrics: