How to Build Meaningful User Segments
Understanding your customers is critical to a successful optimisation strategy. Knowing what motivates some users to purchase, and what prevents others from checking out, is a fundamental requirement to strategic experimentation.
But not all customers are the same, some are impulsive and others are considerate!
This blog sets out to help you find the different audiences that browse your website so you can optimise for them accordingly.
What is a user segment?
A user segment is a distinct set of users that act differently when compared to other users. “Act differently” is important, as there is no point identifying audiences for your website but not being able to act on this information as they all perform the same. Segmenting your users into two groups and targeting them with different experiments only to find out they exhibit the same behaviour all of the time, needlessly increases the length of time it takes to run an experiment.
You also need to ensure that you can identify these user segments online for them to be useful for experimentation. Common personas include data points around income, personality or lifestyle which are useful for adding content and understanding the motivations of a user but I can’t target a personality trait like introvert, or segment my experiment results by this.
Finally, sample sizing is important for experimentation, and my user segments need to be large enough to support analysis to a confident level. You might spot a really interesting trend for users in Bristol that use the first version of Internet Explorer, but if that is only 0.1% of your traffic and will require 500 days to reach a sample size that increases revenue by £50, is it worth your time?
Why do they matter?
By identifying the different types of users that browse your website you understand the different motivations behind conversions and behaviours that these users exhibit. This can then help you enhance the user experience and remove the barriers to conversion for each audience.
For example, if you know that you have a large segment of browsing users that cycle between listing and product pages over and over again without ever purchasing, you might come to the conclusion that there is key information missing on the listing page. This is then an actionable insight that you can use to gather more information through experimentation.
Experimentation will help you to understand the key information your visitors require, in order to commit to a transaction. These learnings impact not only sales, but can increase the efficiency of your marketing efforts as you know which information to include and bring to your customers attention.
Additional insight can be unlocked from existing experiments when breaking down results by previously identified segments too. At its simplest level, splitting results by device can give insights into how user journeys differ from mobile to desktop and give you data on how to improve device specific experiences. When results differ consistently this can also be a clear indication that your on-site experimentation strategy is ready for personalisation.
Inversely, when you’re not seeing differing results through your user segments this can be a sign that there are still gains to be made from traditional A/B testing to a large audience. A common error for marketers to make is to ‘over-personalise’ customer experiences without any data to show that their user base is ready for a customised experience. This usually results in a higher frequency of inconclusive experiments, and the winning tests having a smaller revenue impact than they would if the benefit was served to the entire audience.
How do I find them?
1.What would you do?
The starting point for building user segments should be to think about your own personal experiences when browsing your website and that of your competitors. It’s likely that those experiences aren’t unique to you and are common amongst your user base. At the first level, think about when you visit the website, what devices you use and what your state of mind tends to be.
It’s difficult to template this approach as your user segments should be unique to your website or industry. Every website has new and returning users, but the characteristics of these user types can differ wildly across websites within the same industry. A new user to google maps can have wildly different intentions to a new user visiting citymapper despite there being an argument that the core products are very similar.
2.Do they exist in the data?
The next step is within analytics, where we can check for our 3 audience criteria: identifiable, impactful and showing distinct behaviour. Anything that is identifiable in analytics should be identifiable on the website (unless you are merging 3rd party data after user sessions – but this should be an edge use case).
The most common way to find user segments within analytics is to see how user journeys differ by different visitor properties – i.e. is there a significant difference between how users transition through the website when they come via branded search terms compared to unbranded? How does their journey differ depending on where they are in the customer lifecycle?
It’s important not to overcomplicate your segments to make them look groundbreaking – most aren’t! It may be as simple as new visitors and those on mobile having similar characteristics compared to returning desktop users. Providing you’ve found distinct behaviour and the audiences are large enough to have a real impact on your KPIs, you can begin to align your strategy towards their needs.
3.Give them some life.
Once you’ve found your segments, I find it useful to name them and give them a relevant story. This can help tailor your thinking towards what the customer needs are and align your strategy to their goal. The best experimentation programmes tend to be customer-centric – so your user segments should be too.
Those new visitors and mobile users seen in the graphs above may have a large proportion of traffic but low conversion rates – and looking at the customer lifecycle shows that almost ¾ of revenue is generated after the first session anyway. It’s reasonable to assume that these visitors are researching at this stage, whilst returning users on desktop are much more likely to convert. Labelling these two segments as “researchers” and “buyers” can stop you wasting time trying to make new users convert when they aren’t likely to; instead you can find out what information is important to these and enhance their user experience so they are more likely to return and convert at a later stage.
There you have it! A couple of actionable user segments that bring to life the different ways visitors browse your website.
From this you can stop bombarding researchers with intimidating urgency tactics that frustrate this type of user and instead look to provide them with the core information they need. When they come back, and they will if they’ve had a positive first impression, you know they’re significantly more likely to purchase. That is when conversion tactics can help give the user the nudge they need to get the conversion over the line and turn what may have been another abandoned basket into a loyal customer.
If you’d like to find out more about how you begin to build meaningful user segments for your business, get in touch today!