When you learn mathematics, the first thing you have to learn are the numbers and the operations; with Google Analytics what you have to learn are most basic metrics.
To help you get started you need to understand that:
Tip: If you want to get round this limitation, you can use another web analytics tool which functions using logs.
Exception: you can record the activity of individuals if you can identify the person through their login. If you wish to do so, you must implement a way of identifying the user with a User ID.
Outline of Metrics:
In the following chart you can see an outline of the main metrics.
Imagine that at 9:00 in the morning somebody arrives on your home page via her Google Chrome browser, makes a series of clicks and then leaves. After a while, she returns via the same browser and visits several more pages. A little later, she changes browser and just visits the home page.
We will see the following metrics:
- Time on page
- Duration of session
- Bounce rate
If a user updates or refreshes a page another pageview is generated
In the example on Figure 2 we can count seven pageviews which, when we break down the URL, are recorded as follow:
|Dimension: URL||Metric: Pageviews|
A session is a group of pageviews carried out by the same cookie in a short period of time. It represents the browsing of one person on the website.
How long is a short period of time? Well, it depends: the Google Analytics default setting ends a session after half an hour’s activity, which is to say when no more pageviews are generated. The setting can be changed, as, for a tutorial platform, for example, it may make sense to increase the period of inactivity to an hour.
All sessions must include at least one pageview, and all pageviews must belong to a session.
|Dimension: time||Metric: sessions|
|09:00 a 09:30||1|
|10:10 a 10:30||1|
As you can see in Figure 3, a session may have one pageview or many.
|Dimension: time||Metric: sessions||Metric: Pageviews|
|09:00 a 09:30||1||3|
|10:10 a 10:30||1||3|
It is important to remember that at midnight (00:00) the session is restarted, so if the user continues to browse this counts as a new session. This means a user can have two different sessions within the same half an hour.
Another exception to take into account are campaigns. If a user views a page via a campaign (with an utm_) and returns via another campaign, two sessions are accounted for, though half an hour may not have passed.
A user is a set of sessions carried out over two years by the same cookie. Universal Analytics interprets each cookie as a user. Every time a person visits a website they are assigned a cookie (a unique code which identifies them) which is kept by the browser. When the page is accessed on return visits, the browser lets Universal Analytics know of the cookie assigned to it on the previous visit. In that way it is possible to monitor over time the activity of the user on the same website.
However, if a user switches browser or deletes cookies, sessions are not connected so a new cookie is assigned.
Google keeps cookies for two years, so if a user visits a website with the same browser after two years since the last visit, the person is considered a new user.
In Figure 4 we can see the breakdown of numbers of users by browser
|Dimension: Browser||Metric: Users|
So a user must have at least one session, which in turn, must have at least one pageview. And all pageviews must belong to a session which, in turn, must belong to a user.
If we carry out an analysis by hour, we would have:
|Dimension: time||Metric: users||Metric: sessions||Metric: Pageviews|
|09:00 a 09:30||1||1||3|
|10:10 a 10:30||1||3|
An as you will have worked out by now:
The number of users is less than that of sessions and the number of sessions is less than that of pageviews.
Time on page
The time on page is the number of seconds that a user takes to make a click linking us to another of our pages during the same session. Very important:
The time on page is not the real time which the user is on the page.
What Universal Analytics does is take the time on which the pageview is produced and it calculates the difference between that time and the time when a new page is opened. An example will make it easier to grasp:
The user who visits the page at 09:00 generates a pageview and after 20 minutes goes to our blog. Universal Analytics simply deducts the first time from the second:
09:00 – 09:20 = 20 minutes
So according to this the user has been 20 minutes on our home page, which is equivalent to 1200 seconds (20 minutes x 60 seconds)
But what happens when we don’t have a second page? What happens to the last case in Figure 5 for example? Well, it simply isn’t calculated. So,
If the user only sees one page, the time on page is 0 seconds.
The time on the page is underestimated if the user only sees one page, or it can be over-estimated if the user leaves the page open, carries out other tasks and then clicks on one of our links.
So, following the example of Figure 5, the times on the page would be:
|Dimension: URL||Metric: Time on page (minutes)||Total Time on Page (Minutes)||Average time on page (Minutes)|
|Home (1)||09:20 – 09:00 = 00:20
+00:00 – 09:30 = 00:00
+10:20 – 10:10 = 00:10
+00:00 – 11:10 = 00:00
|00:30||00:30/4 = 00:07:30|
|Blog (2)||09:30 – 09:20 = 00:10||00:10||00:10/1 = 00:10|
|Product Tab (3)||10:30 – 10:20 = 00:10
00:00 – 10:30 = 00:00
|00:10||00:10/2 = 00:05|
Duration of session
The duration of a session is the sum of all the times on page. It’s as simple as that. We take all the pageviews belonging to a session and calculate the differences in time between the last pageview and the first.
So the session times would be calculated as followed.
|Dimension: session||Duration of session (Minutes)|
|Session a||09:30 – 09:00 = 00:30|
|Session b||10:30 – 10:10 = 00:20|
|Session c||00:00 – 11:10 = 00:00|
As you can see, the same criteria are used as for calculating the time on page. With the same consequences for overestimating the time when the user is doing other tasks between pageviews, or underestimating when there is only one pageview the entire session.
It is possible to obtain the average session by user:
|Dimension: user||Duration of session (Minutes)|
The bounce rate is a complex metric which depends on the dimension and the formula applied.
There are two types of bounce rate: by page or by session
a) Bounce rate by page:
Imagine that a user accesses our home page, sees a distribution page, then leaves without looking at more pages. That is a bounce. The bounce rate, then, is the number of bounces out of the total number of pageviews.
Bounce rate = number of bounces/total number of pages
The bounce rate tells us how well the page has worked. If we consider that our aim is for users to continue using our website, this metric is particularly important for landings and distribution pages.
The lower the bounce rate, the better
In the figure, the bounce rate by page would be as follows:
|Dimension: URL||Metric: Bounces||Metric: Bounce rate|
|Home (1)||3||3/4 = 75%|
|Blog (2)||0||0/1 = 0%|
|Product tab (3)||1||1/2 = 50%|
|Total:||4||4/7 = 57%|
A 100% bounce rate per page would mean an average time on the page of 0 seconds.
b) Bounce rate by seconds:
Imagine that a user goes to our home page and leaves again without looking at any more pages. This is registered as a bounce session. In this dimension, the bounce rate accounts for those sessions which only have one pageview. As in the other dimension:
The lower the bounce rate, the better
So, in order to calculate the session bounce rate we have the following:
|Dimension: session||Metric: Bounce||Metric: bounce rate|
|Session a||0||0/1 = 0%|
|Session b||0||0/1 = 0%|
|Session c||1||1/1 = 100%|
|Average bounce rate:||1/3 = 33%||100%/3 = 33%|
As one can see, one of the sessions has only had one pageview, representing 33% of sessions.
A 100% bounce rate means an average session duration of 0 sessions.
I am conscious that some exceptions have not been indicated and that I have generalized, but with this small guide, you can interpret 80% of metrics correctly. You can find more information on the Universal Analytics website.
Google Analytics for developers: This is ideal for seeing specific details about the tool.
Analytics Academy: This is a course run by Google to teach the basics of Google Analytics.
Written by Federico Ferreyra