Segmentation in Google Analytics

Segments are a way to isolate sessions or users based on their inherent characteristics. 

When you look towards the top of the page in Google Analytics you can see an area that points to the segment of the population that you are looking at. It usually looks like this picture below:

Once you click on +Add Segment, you can choose from a list of “All segments” that exist for this email account. Here is a picture of the account I am using.

Segments work entirely based on email accounts and are linked to them only. That means that you can create a segment for one view and can use it for any other view too. An important point to note is that ONLY the definition of the segment is used across different views or properties. No data gets passed along with the segment definitions.

As you can see in the picture below, there are pre-set “system” segments available to you in Google Analytics. They are mostly defined by traffic sources like “Direct Traffic” and behavior like “Bounced Sessions”. These are already available, and you can select any number of them to compare the behavior of each segment side by side. 

In the image below, I have selected three different segments here.

You can select up to 100 segments per user but from my experience, any selections beyond 4 segments make the reports very cluttered and not helpful. I would always recommend comparing mutually exclusive segments like the traffic types above for best analysis.

There are also several other types of segments. They are as follows: 

1. Custom Segments: These are the segments that a user can build based on their own dataset. The definition of these segments can be shared across several accounts.
2. Shared Segments: These segments are created by some other account and shared with this email account.
3. Starred Segments: The segments that are used most often can be “starred” and be available in this short-list. This is great for people who have created hundreds of custom segments in their accounts.
4. Selected Segments: If one has a list of several different segments selected from the custom segments, this is a quick way to see all of them in one place.

Building Custom Segments

The best way to create a custom segment is to click on the red button “+NEW SEGMENT”. This is the screen you will see after clicking through.

For the sake of explanation, I have marked the areas into three groups. Group A marks the different ways in which you can create segments. Group B marks the choices that you can make to define the segment. Group C lets you know how much of the total data is captured by this segment of the dataset. The default is Demographics as seen above but it can also be other classifications. They are shown below:

Technology: This is very important to understand how the site performs from a technical perspective. It is particularly useful for specific mobile devices and browsers.

Behavior: This answers the questions on how many sessions and transactions per session.

Date of session: We can change the dates on the top menu but when you want to compare cohorts based on dates or months, this segment becomes important. Think of LTV.

Traffic sources based on UTM parameters. This is specifically used for seeing traffic based on several campaigns simultaneously.

Conditions based on Dimensional Variables. This is used most often because we can create segments based on any number of dimensions and their values; and group them together. It becomes very slow if using more than 5-6 And/Or conditions.

Sequence of Steps in a user’s path based on Dimensional variables: This is particularly used for UX and product analyses where we want to explore the behavior of users based on calls to action that are steps towards a goal.

Segments are one of the most powerful sections of Google Analytics and when used well, can result in the mining of vast data riches. Often, I see a rush to build views where segments will work better and require less administration.

Bibliography on Segments:

1. Google Analytics Documentation

2. Loves Data Tutorial.

3. Creating in-depth segments.