At Springboard, we’re privileged to sit at the unique intersection between data science and UX design. Having spoken with hundreds of industry experts in both fields, we’ve come to believe that there’s a lot of common ground between the 2 fields. Data scientists check user behaviors through analytics tools, while UX designers look to see what best suits the large numbers of users they serve.
It’s a pity that UX designers and data scientists rarely talk — they can learn a lot from each other. Content that combines insights from data science to UX design or vice-versa is immensely valuable, and scarcer than we like.
This led us to think about the data science principles that are helpful for UX designers, and ways you could implement these rules into your day-to-day workflow. So here’s what we came up with.
Embrace metrics and think of the return on investment
Most UX designers aspire to build something better for their users.
While that’s a noble goal, how do you determine if something is truly “better”? Metrics can help you align closer with business goals and showcase the return on investment your company or client gets from your efforts.
The metrics that businesses care about vary. In some cases it’s revenue. In others, it’s on-page conversion, time spent on a page, or number of photos uploaded.
By embracing the metrics your business cares about and contextualizing why good UX design matters to those metrics, you not only make it easier for business leaders to justify putting more resources into UX, you also start prioritizing what initiatives make the most sense. Speaking to how better user engagement ties to company goals helps frame your work in a context that’s familiar to non-designers.
How you can apply this to your day-to-day: Ask which metrics matter to your company or client. Deeply examine how your design goals will help improve those metrics and the company’s bottom line. Then reconcile your work with the impact it drives on the bottom line.
Key resources: UX designers working in data-driven organizations should start by reading Dave McClure’s ARRRR Startup Metrics framework to understand how companies measure growth at various stages. The book Lean Analytics is a must-read to understand how good use of data works to track impact.
Experiment, measure, and repeat
You’ll often find yourself debating between many design treatments. How do you make a decision?
Good news: You don’t always have to decide — you can let users tell you.And to do that, you’ll want to take a page from the data scientist’s approach of experimentation.
Data scientists tend to think in terms of experiments — they’re usually pretty structured about stating their hypothesis for an experiment, what they wish to measure and learn, and how they want to run the experiment.
This mentality might already be part of your workflow, but you should employ some tricks data scientists use to make sure they’re on track with their experiments. State and record your hypotheses. For instance, “I believe that version 2 will lead to 20% more users making a purchase than version one.”
Next, build the minimum viable version of each design.
Then put your design in front of users and track their behavior to learn if your hypothesis is true. Make sure to consider the concept of statistical significance (whether you have enough observations to make a statistically accurate prediction).
How you can apply this to your day-to-day: Get familiar with deciding when to incorporate data in your decisions, and when data doesn’t matter. Learn how to use tools like Optimizely, Mixpanel, or Google Analytics that can easily help you see how your UX choices impact user behavior.
Whenever you’re thinking of testing a new design idea, record your hypotheses. For example, if you want to change the color scheme and typography to make instructional content easier to read, say that you’re testing to see if making instructions easier to read in a certain way leads to more user engagement.
Key resources: Here are some quick tips on how you can get started with Google Analytics. And here’s a recent InVision Blog article on A/B testing, a form of experimentation that can help you determine what version of a webpage performs best for certain business goals.
Segment your users and learn how they’re different
You’re likely already creating user personas that map onto the most common profiles of a company’s users.
Broad quantitative data from tools such as Google Analytics, which captures demographic, behavioral, and device data, can help flesh out those user personas in a way that’s richer and ties more closely with users’ actual behavior.
With Google Analytics, you can quickly determine that older users may have different expectations from your site than younger ones. Males and females may respond differently to certain color schemes. You can even get a rough view of what interests your users have — and how they might react differently accordingly.
Don’t stop at explicit data, either. Data scientists often infer data with clever techniques. They might map user first names to a database of the most common female and male first names to come up with a pretty precise picture of the gender split in a dataset. If there’s data you need to flesh out for a persona, think about how you could get it, or work with technical team members to get it done.
How you can apply this to your day-to-day: Ask about quantitative data that can help define which of your personas are the most common and which can flesh out more of who they are. Be thorough about who you’re designing for, and learn how to segment your users. Ask for as much data as possible, any surveys people have collected around your target groups, and look as deeply as possible about who your users really are with a mix of interviews and a look at the quantitative data.
Key resources: Read about how Google Analytics can supplement your audience data so you know what’s possible, and read about what data scientists think are the most important ways you can segment users.
Personalize at scale
You may have heard the term machine learning. One thing that machine learning can enable is a personalized experience to as many people as possible with the power of automation and algorithms.
How does Amazon know exactly which book you might like to read next? It’s through the power of machine learning algorithms. By using these algorithms to infer users’ preferences (e.g. assign a probability to how likely you are to find a book enjoyable), Amazon delights users by consistently providing them with something the company knows they’ll be interested in.
Don’t lose a chance to delight your users by personalizing experiences and storytelling with the data they give you.
Data scientists apply algorithms at scale to ensure that every user gets a personalized experience. They use complex mathematics and programming to do that. You can capture some of that magic by using tools like Visual Website Optimizer to offer different user experiences for different groups. The classic example of this is localization: Having your application load in a user’s preferred language can mean a world of difference for them.
How to implement this in your day-to-day: Understand how user experiences can be personalized and where the biggest benefits are. Use your understanding of different user groups and personas to build experiments so that every user walks away delighted.
Get and understand feedback at scale
You’ll likely rely heavily on one-on-one interviews and focus groups. While these are great at providing a rich context of how a few users feel, you can supplement that qualitative feedback with quantitative sources that measure users at scale.
For instance, you can use user surveys to supplement your user research interviews. Similarly, you can use actual usage data to supplement user testing interviews.
By combining both deep qualitative and quantitative insights about how users are truly interacting with your product, you can get a complete view of how users are reacting to your design — and how you can improve it.
How to implement this in your day-to-day: Use tools like Qualaroo to ask users questions when they’re visiting your site. Use Hotjar to track heatmaps and on-page analytics of how your users move through your app. Use tools like Promoter.io to get an overall view of how users like your product at scale.
Key resources: Read about the one number you need to grow, the Harvard Business Review article that introduced Net Promoter Score as the gold standard for measuring user satisfaction. Then delve into how you can analyze and visualize user feedback.
Think of data as a tool you can use to sharpen your instincts about your users and what they like. It’s a superpower that can help you create rich, immersive, and personalized experiences that will be maximally useful for your users.
Suhail Doshi, the young founder of analytics powerhouse Mixpanel, once opined in a New Yorker feature that “most of the world will make decisions by either guessing or using their gut. They will be either lucky or wrong.”
Your commitment to using data to delight your users is an utter commitment to being right rather than lucky.