I’m a data alchemist at Lever, which means I use principles of data science and design to create value from data. Hiring is a complicated process, involving a lot of people doing a lot of things. I apply my passion for finding patterns in data by making sure we collect and analyze all the things happening inside Lever. In order to deliver a powerful recruiting reports tool that cut through all the noise, we held ourselves to a few essential data visualization and design principles to create Lever Reports.
1. To build something that people want, start with throwing away your assumptions.
Simply put, if you build a tool based on your assumptions, you’ll end up with something that provides value to only a few people. With Lever Reports, we knew we needed to give people a way to understand the decisions they make, and show their decisions in aggregate. But we also knew we made many assumptions about how people use our app and what they want to see about how they hire.
You need clues to determine the best solution to build. User research will give you those clues.
There are a lot of ways to divide a number by another. A lot of ways to calculate a percentage. But it wasn’t until we talked to a lot of customers that we understood what people were trying to measure and understand. User research allowed us to accomplish 3 things:
- Confirm or deny our assumptions.
- Understand the underlying problems, and its impacts on the team.
- Confidently prioritize the elements we needed to build.
And there doesn't need to be an end point to user research. We interviewed customers from the early design stages all the way through QA testing. We released the reports in phases, went back to customers for feedback, and shared mocks of the upcoming reports. There’s simply no substitute for learning directly from users.
2. Trust will ultimately determine whether people will use what you’ve built. Transparency is a necessary ingredient for trust.
In the previous version of Lever Reports, all we showed were the final calculations. But the problem was, sometimes the numbers would surprise people and we had no way of showing how we came up with that number. You see, the app can’t capture reality exactly because it’s prone to human error -- at some point, a person has to update an interview stage or input information into Lever.
In the new reports, we made a hard rule: for any number that we show, you should be able to count yourself in a list of candidates that we used to calculate the number. More often than not, you need to earn customers’ trust. This is especially true when the value you offer is the data you collect for customers.
3. Your product shouldn’t hold customer data hostage.
Data brings awareness. Being able to see patterns and gain broader understanding often only becomes available when you collect good data. I’ve come to see over time that if you collect data about different processes, you can gain new insights into what, why, and how things are happening. (Especially the kinds of things that are pretty much impossible to see from just anecdotal information.)
The data we collect at Lever reveals how our customers are doing and what they could be doing better. While it’s not the entire story, data does bring objectivity -- and with that clarity -- to the table as teams uncover places where things aren’t going as well as they could be.
To fit our customers’ needs, we built layers of specificity into our reports. And then we built a filter tool to unlock the data treasure trove that housed the answers to questions like:
- “How does Annie’s interview feedback compare to that of the broader engineering team?”
- “How long does it take us to hire a candidate for the iOS Engineer role from the time that a candidate enters our system?”
- “How many interviews did Keith schedule last month for passive candidates who we sourced from LinkedIn?”
4. The way things are ordered makes a difference. Information hierarchy is what makes reports powerful (or weaksauce).
It’s an obviously simple rule to follow - the important things are big at the top and easy to read. Below that, more detailed information is in smaller letters or tool tips. This allows people to get what they need quickly. And for those who need more information, they’ll take the additional step to hover over interesting numbers or scroll down to learn more.
Good data visualization is critical here because there are a million ways to slice and dice data. We chose four to five criteria that we know everybody cares about. So whether you have a team of 2 or 200, our reports overview pages show you essential relevant statistics. And we created a “hero statement” at the top of each report so people could quickly learn what’s happening in their pipeline. Simply put, our reports bring your data to life. Each report tells a story that tell our customers how to improve their hiring process; sometimes with small tweaks and, when necessary, with drastic changes.