It’s the most magical time of the year… to look back at your product analytics and uncover new insights. A hot drink, a quiet office, and a full calendar year of data that you can turn into tidy, pretty charts. This is truly the best present for a growth nerd.
We’re going to talk about segmentation as a data analysis prompt to uncover surprising insights. Slicing and dicing your use base can be as satisfying and less traumatizing than other office holiday traditions (looking at you, cringe-worthy Zoom parties!)
Your community leads,
Ran Liu, Molly Norris Walker, Chase DeNomme and Scott Christensen
What is segmentation?
Segmentation is a super important and useful form of data analysis you can do to uncover unexpected growth insights. For example, when you look at the metrics you're trying to move, you can cut or segment based on different criteria, such as:
- Role/Job title
- Product usage
- Time in product
Common segmentation frameworks
Chase’s favorite framework represents different stages of the customer lifecycle:
- Strangers - I’m not yet familiar with or aware of a particular product, brand, or service.
- Evaluators - I’m in the process of considering the offering and deciding whether it aligns with what they're looking for.
- Beginners - I’m someone who has recently started using the product or service.
- Champions - I’m the most loyal and satisfied customer who has extensively used the product, understands its value proposition, and derives significant benefits from it.
Probably the most popular framework used in growth looks at time-based cohorts of users and their product usage:
- New users - recent joiners
- Core users - central, consistent user base that regularly engages with the product
- Power users - a subset of highly-engaged users pushing the boundaries of what the product can offer.
Although common, these are fairly advanced frameworks. Don’t be afraid to jump into self-serve product analytics tools you might have in-house to see what insights you can gain from more basic out-of-the-box segmentation criteria such as geography or device.
3 examples of surprising segmentation findings
- At Amplitude, Ran redesigned the signup experience from a public product demo available from the marketing website. When she segmented by mobile visitors, she saw the mobile conversion rate actually dropped with the design changes. Uh-oh! She found a mistake where the sign-up module on the demo wasn’t properly optimized for a mobile sized screen and was being cut off.
- At Expedia, Scott segmented a business traveler onboarding experience by geography and found that European travelers were more likely to add passport data than US users who traveled more domestically. He removed the passport step for US-based travelers and saw higher onboarding completion rates.
- At InfluxData, Molly segmented power users by feature usage. She discovered that power users with the highest product consumption all adopted one key feature. She then reworked the user experience to promote that feature and expanded product usage and revenue.
Let’s keep it real--Most of the time you do segmentation and nothing particularly interesting will show up. So it does require trial and error, and maybe even luck to find an insight.
Growth is hard. Pushing metrics is hard.
Even if you find something interesting with a segmentation exercise it could be a weak correlation and not necessarily a causation. So as a designer, it's very, very important for us to go back to user research and qualitative methods to validate what you're inferring is true before you do any experimentation work going forward.