Unravelling User Behaviour Segments
An understanding of who the people are who use your digital interfaces and how they use it will go along way to optimising experiences. Yet, obtaining this level of insight from quantitative data takes a bit of practice.
Glance at a digital analytics dashboard and you will most likely see visitor volumes and conversions. This is great for showing your progression, but it’s not telling you how you can improve on a user level.
We know that each person behaves in their own unique way for different reasons.When it comes to assessing individuals using quantitative data though, the count of variables and contexts are too great to assess. Not to mention, analysing individual behaviour will take some time.
This is where audience behaviour segments come into play. Segments allow you to group together behaviour trends and patterns. With this granularity of data we can begin to understand behaviours on a greater level.
We can define a segment as a data set filtered by dimensions or metrics. A combination of variances can be applied to create something more complex. They can be used to assess data retrospectively which saves the need to create a new view or wait for data to accumulate.
Let’s start with a few segment examples:
Campaign source sessions
Campaign does not match (not set)
Users arriving at the site from campaign sources will behave differently. Conversion, devices, and content reports will be affected by the origin of traffic. Using this segment will help you understand what issues are arising for this audience.
Mobile sessions starting from the homepage
Session Device Type = Mobile
Landing Page = Homepage
Say we designed a user journey for mobile devices whereby landing on the homepage would influence behaviour. We would want to cut out noise from journeys starting at specific landing pages.
Segmenting by landing page is important. Users will behave differently having seen a particular set of content from the start of their journey.
User reached a product detail page from a specific listing page
Step 1: Page contains summer_sale
is followed by…
Step 2: Page matches regex products\/.[a-z]
This sequence condition will allow you to assess user journeys that follow a particular path. In this instance, we’re highlighting a flow between a sales page followed by a product detail page. The regular expression will match any page that is in a sub-folder of /products but not the /products page itself.
Regex will enable you to deploy more advanced segments, so it’s certainly worth checking out this starter guide.
Segments to test hypothesis
A heuristic analysis or user feedback session may identify a pain point or a barrier to completing a conversion. Applying a segment that reflects this can help validate if this issue is apparent in aggregated data.
If a particular page appears to be important for users that do convert — create a segment to analyse the behaviour of users that view that page.
If there is an element on a page that is below the fold that has been identified as problematic, use scroll depth tracking to create an event. This event could then be used to create a segment.
The example below highlights users who visit the shipping page and scrolls down to view at least 75% of the page.
Event action = 75%
Event label = Shipping
Use edge cases
More often than not, we consider outliers in data as anomalies that need to be filtered.
What if we could group together extremities to find patterns that could be used to optimise experience for all users?
Examples of edge cases should appear when you analyse your user behaviour. When grouped, these users will add up to a small percentage of total users. Instances of edge cases can include users who:
- Interacted with an under utilised feature
- Returned to the site very frequently
- Were very engaged with content but didn’t convert
- Took many days to convert
- Bounced from the site (While this won’t necessarily be an outlier in any case, this group of users are at the opposing end of the interaction spectrum)
When these ‘edge case’ users are grouped together we can try and determine what caused them to behave in this way. Out of this, we could attempt to amplify positive circumstances or improve negative experiences.
Getting the best results
Segmenting data will take you a step closer to understanding who is using you site in what way. To get the best out of them try to utilise qualitative data to guide what segments you should be deploying.
This combination will enable you to ask the right questions about your users and find suitable answers.
Lastly, don’t close the book too soon. The relevancy of segments can change according to behaviour trends. Keep on top of what data needs grouping together. It may take time to identify the best segment to find the most useful insights.