In product design, constraints are good. They help us assess trade-offs and confidently make the best decisions. Ultimately, constraints create more targeted, more optimized experiences. Through analyzing data, we can create constraints where previously there weren’t any.
For the sake of this post I’m going to use a specific example of a project I worked on last year : Gusto Job Offer Letters. The problem I was tackling was designing an experience for potential employees of our customers.
At the beginning stages of the project I was trying to understand what information people expected as a part of a job offer. We wanted to make the offer letters really easy to understand—but also excite and surprise the employee, making a great first impression.
If there aren’t enough constraints, make some.
So how do you use data to help your design? Start by asking some questions:
What are we uncertain of? What might we need to know to make a decision about that uncertainty?
To make the offer letter personal, I wanted to include a photo of their manager and a personal note (you know, fancy stuff…) but wasn’t sure about the hierarchy of those elements. I was thinking: What should we prioritize in the first version? How often would there be a photo to show? Can the design degrade gracefully?
A common problem of designing software for business is falling into the trap of designing for the ‘ideal’ state, when an account is 100% set up and already using all of your features.
Our data team ran some queries on our target audience and found that, of the companies that hired employees within the past four months, only 40% of those employees were assigned a manager. And of those managers, only 5% had a photo.
This means less than 2% of any given offer letter would include a manager’s photo. Which means—you guessed it—the design shouldn’t focus on highlighting the manager’s face! We avoided creating a consistently awkward experience by not letting this element be a focal point of the design.
Creating constraints with data increases confidence in your decision making and leads to better design.
The data provided us with some much needed constraint, yes. But the real value of that constraint is focus. If our job is to solve problems with design, data helps us narrow in on what those problems are, making the path to a solution much clearer.
I wanted to keep going, so I asked another question:
What screen sizes should we aim to design for?
We knew what information to include — but how should we order it? How wide should the letter be? Do we really need to make it work flawlessly on mobile for the first version?
Google Analytics is one helpful tool for analyzing data — but what do you do with a table of 50,000 entries? Enter the Data Science team.
If you don’t have a Data Science team, you just need to ask a favor from a software engineer. Cherish this relationship. A knowledgeable engineering partner who knows their way around your database be a query-running lifesaver. (Or if you’re feeling adventurous, you could learn to run them yourself!)
We took the data and segmented by user type (i.e. employee, contractor, admin, accountant). That’s when things got interesting.
First, we segmented into our different types of users: employees, contractors, admins, and accountants. Then, we were able to cluster people who logged into our app by device: smartphone, laptop, high-res laptop, desktop monitor, tablet. We have the upper and lower bounds of each, as well as the average height and width of each cluster.
In the end we were able to discover some really helpful information:
- how many employees log in using mobile
- the minimum pixel width that covers 98% of our users for each screen category (mobile, tablet, browser)
Which resulted in a great design consistently across devices :
Takeaway for product designers: If you don’t have enough constraints to confidently move forward, create some by using data. Also, if you have enough data, you can check if a problem is worth spending your time on, or if you’re reaching diminishing returns.
Takeaway for data scientists : You can empower all types of teams to do better, more focused, more targeted work. You’d be surprised how useful your data can be for so many different types of people at your company!