Data-driven design is a fancy phrase these days, but what exactly does it mean to design with data?
Does it mean to perform more A/B tests, or conduct more usability studies? Turns out, data-driven design is much more than that.
In this article, we’ll cover what data driven design is and whether data actually allows us to make better design decisions.
We’ll also talk about common misconceptions in data driven design and cover techniques to gather data that can be used to create successful digital products.
Let’s dive in.
Data-driven design encompasses techniques and approaches to gathering and interpreting data to make better design decisions. Data driven design can be applied to any design process, but most often is utilized in a digital environment where it’s easier to log user behavior data and use this data to inform or guide your product design.
As an example, the team at Google employed rigorous testing data to determine which of the 41 shades of blue would perform better in their interfaces before finally committing to one.
But the term “data-driven design” is a bit confusing as nowadays it encompasses three different approaches to how you can use data when designing your products.
There is data-driven design, data-informed design, and data-aware design.
Let’s briefly discuss every approach to sort out the confusion.
You are data driven when data determines your design decisions.
Imagine you are testing two prototypes and the first prototype has a 10% higher task completion rate. When you are driven by data, you should be ready to opt in for the first prototype based solely on the data you’ve gathered.
Important: when data plays such a major role in your decision making process, you need to apply extra effort to obtain high-quality data. Otherwise your team will be making poor design decisions.
As an example, if you test your product with users not from your target audience, chances are you will be misguided in your decisions.
Note: the team at Netflix vastly uses data driven design in their process. We’ll cover their workflow and other examples later in the article.
You are data informed when data is a part of your design thinking process. In other words, you use data to improve your design solutions, not determine them.
Let’s take the aforementioned example where users completed 10% more tasks when interacting with one of the prototypes. Data-informed design team takes this piece of information and explores the issue further. Turns out, the users that completed more tasks, were less happier when completing them. On the other hand, the marketing team shared that conversion rate was equal for both prototypes and the completion rate didn't matter that much.
Now the design team has several things to discuss and the data gathered so far is only a part of their discussion.
The data-informed design team uses data to explore design challenges further and tries to understand what else can be changed to improve the product.
When you employ a data aware approach, data becomes a design problem itself. In other words, you are aware that data can be collected and used in multiple different ways to test many different assumptions. You start experimenting with what data you can collect to solve your problem.
Back to the example with a 10% higher completion rate prototype: being data-aware, you realize that there can be different ways to measure the completion rate. Were you tracking people automatically or were you asking them in a survey? Were users somehow prepped before completing the tasks? Would the completion rate be the same if users interacted with the whole product, not just a part that you were testing?
Put simply, with a data aware approach, the way you collect data becomes part of your design solution.
All three approaches to designing with data are closely related: often designers use a combination of all three when solving design challenges.
But knowing the difference is important -- when a particular approach is applied on a bigger scale, it completely transforms how your team designs products.
Now that we’ve discussed different approaches to designing with data, let’s cover how data-driven design can be used to create high-performance websites.
One of the common uses of data in design is the process of creating high-performance websites and services.
Let’s briefly cover what a data driven website is, and then explore how different companies use data to create successful digital products.
Data-driven website is a website that is designed or updated with every design decision being driven or informed by data to maximize the website’s commercial performance.
Even small changes to design and layout can significantly change how your customers interact with your website and applications.
But the way you use data to improve a website's performance highly depends on the type of your website.
For e-commerce websites, simple UI changes such as modifying buttons or colour played insignificant roles in their financial performance, according to a Qubit study of 6700 ecommerce experiments.
For example, modifying website buttons produced incidental 0.2% fall in revenue, whereas changing color didn’t affect revenue at all.
On the other side, e-commerce websites benefit greatly from reevaluating their UX patterns. A redesigned product recommendations block produced 0.4% uplift in revenue, whereas informing users of others’ behavior and stock pointers together increased revenue by more than 5%.
Using data, a team at Smartwool discovered that users scan the website better when all the image products are structured in the same manner.
After they optimized the layout for product pages, data backed up their initial hunch: the new design led to 17% increase in revenue per visitor.
Unlike with e-commerce websites, color and buttons can make all the difference for a digital service.
While all e-commerce websites are more or less using the same UX patterns, every online service, be it Saas or a digital app, is unique and often contains exceptional behavioral patterns.
That means even small design tweaks may affect user behavior in a big way.
That’s why relying on data insights becomes critical when you optimize digital applications.
A long time ago a product team at Spotify realized how crucial the data driven design was and since then have been using data extensively to test their assumptions.
The team A/B tested two different versions of a website: light and dark. The data demonstrated that users on average spent more time listening to the music with the darker team.
For a music streaming giant with over 300 monthly users, even adding a couple of seconds to a listening time made the redesign worth the effort.
Later the team would test with 5% of their users whether the interface should have play buttons.
Data showed that users were more prone to play songs and playlists with no buttons at all.
We’ve heard many times about how data driven Netflix overall as a company. Whether it's recommending what movies to watch, deciding what original projects to fund, and what TV shows to continue, Netflix hugely relies on analytics and data gathered about user behavior.
Turns out the very same data driven approach is used when it comes to designing as well:
For example, Netflix designs movie posters and previews solely based on user interaction data with no subjective input from designers at all.
So whereas Spotify is leaning towards a data informed approach, Netflix could be described as guided by data.
There are several techniques that allow us to gather data that can inform or guide your design process.
Let’s briefly cover the most common ones.
User testing allows you to gauge opinions from your target audience. There are two approaches to conducting user tests: moderated user testing and unmoderated user testing.
Moderated user testing. This type of user testing is usually performed with a small segment of users. The main challenge here is to ensure that every user performs the same set of tasks in the identical environment, otherwise it’s hard to render results of such tests as objective data.
Unmoderated user testing. This type of testing is usually performed with no moderator present, which allows to quickly gather and test the product with a larger segment of user database.
All in all user testing is better suited for gauging user preferences and exploring design theories. It’s always better to combine user testing with other methods of gathering data to test your insights.
A/B testing is one of the most popular techniques for obtaining data about how users interact with your website. Put simply, with A/B testing you obtain data about how users from the same target audience perform the same tasks with different versions of your product.
A/B tests are usually performed with a large number of people to obtain reliable data about user performance.
As mentioned before, Spotify used A/B testing extensively to determine how well different versions of design perform compared to each other.
There are already web services that contain large sets of visual data that can be analysed to inform certain design decisions.
For example, services such as Dribbble and Behance both have trending sections where you can easily track what artwork is the most popular right now. Designers can take it one step further and spot current design trends in terms of most popular colors, design styles, and interface patterns.
Although this data should not be used to guide your design decisions, it helps you to discover industry context for your design work and inspire you to test certain concepts yourself.
Although global brands have the luxury of using tremendous amounts of data to guide their design process, even smaller companies start to realize how crucial data is to creating successful digital products.
Every year more and more tools allow us to gather more high-quality data about how users interact with our products.
We just need to learn how to fight our biases and get the most out of the data we have.
At Lucky Duck we’re strong proponents of using several techniques at once to gather data when it comes to designing and improving world-leading products.
If you want to learn more how our team is using data insights to inform and improve our design decisions, feel free to sign up for our newsletter.