With a goal of creating the most developer-friendly cloud computing products in the market, it’s only natural DigitalOcean would be equally invested in building a developer-friendly workplace. The latter is where VP of People Matt Hoffman and VP of Engineering Greg Warden collaborate.
Evaluating performance for highly subjective roles such as engineering and design has always been a challenge. There’s no universally agreed upon definition of “good code,” especially in a collaborative, fast-paced startup. Since joining the team in 2015, Hoffman has worked to navigate these potential ambiguities by building a feedback-focused performance evaluation framework that relies on as much real-time feedback from multiple stakeholders as possible to get a holistic view of Engineers’ performance.
In addition to leading DigitalOcean’s distributed engineering teams, Warden has collaborated with Hoffman to create a framework to untie value and performance, set better goals for engineers and managers, and continue to develop an idea- and experience-driven culture.
In this exclusive interview, Hoffman and Warden share how DigitalOcean defines success for engineers and personalizes employee development, and why moving away from metric-driven performance evaluations helps reduce bias.
Build a framework for innovation
Hoffman described his core responsibility as “to create the philosophical underpinnings of how we think about performance and development, and what growth looks like for both employees and the company.” Before he joined the team, the company was fewer than 100 people and “you’d sit down with your manager once a year or so and kinda figure out how you were doing.” With 400+ employees across the globe, that luxury no longer exists.
“You can do that when you’re small and everyone knows each other,” Hoffman says. “As you scale, that’s not sustainable, and you want to put in more rigor around the developmental focus. You can drive significantly more performance by clearly setting goals and expectations, and then supporting and empowering people to get the job done. We make sure our performance conversations are driving growth by focusing on what people are going to do in the future instead of just what you did in the past.”
Hoffman and Warden broke down DigitalOcean’s goal setting and performance management process into four principles:
1. Challenging but realistic expectations: While DigitalOcean is always raising the bar on performance, an effective goal-setting process starts with setting upfront, experience-appropriate, outcome-focused expectations. “The hardest thing for new managers is having a good understanding of what expectations are for a given role,” says Warden. “For me to set goals for you, I have to know what good goals for you are!”
“A junior person could do great work beyond what your expectations were, but that might not look very impressive when compared to what your Principal Architect does,” he says. “And that Principal Architect could have done 10 times as many things as that junior person, but your evaluation might be skewed because your expectations should’ve been even higher.”
2. Structured goals with short time horizons: “Goals have to be structured because if they change all the time, you can’t actually give feedback,” says Hoffman. “You want to be clear on what’s needed, and you want the time horizon to be short enough that you can be iterative. This allows for much more flexibility and agility across the process.”
Setting year-long employee goals is a rigid and antiquated process. “Instead, say ‘here’s what I want you to work on this quarter, and then let’s see if that’s the right thing next quarter.’ Let’s measure it, see what worked well and what didn’t, and that can inform how we set expectations and focus for the next sprint cycle.”
3. Real-time feedback: “Feedback is in real-time, because even in three months so much can change”, says Hoffman. “A company like DigitalOcean is moving so quickly that we don’t have the luxury of waiting for formal feedback windows. We encourage our employees to always be seeking out, and giving, radically candid feedback.” Real-time feedback allows for faster adjustments, increased engagement, and helps prevent projects from going awry.
“You can have a very productive team that’s creating code and shipping all the time, but they got the product wrong,” says Warden. “That’s the worst outcome. There could be another team that’s slightly slower and more methodical, but they build the right thing. You would rather have that. It’s not black and white ever, but you need to think about more than just production.”
4. Recalibrations: In addition to continuous feedback, DigitalOcean also conducts more formal performance check-ins once or twice per year. This time is used to make sure employees within each role are being evaluated according to similar expectations and also provides an opportunity to ensure high performers are being paid equitably compared to their colleagues and the market.
“The idea of compensation isn’t just to measure performance,” says Hoffman. It’s to make sure you’re meeting market needs and paying equitably, and creating a situation where people aren’t constantly looking for jobs because they feel they are not being paid fairly. We want to make sure our best people feel really well taken care of and recognized for the impact they are creating. By creating agile cycles around performance feedback, we also have more real-time data against which to evaluate compensation.”
Another core function of the formal performance calibrations is to deliver 360° feedback. By sourcing feedback from peers, stakeholders, managers, and collaborators, you can gain a more holistic understanding of an employee’s performance.
While differentiating performance is easy to do on average, it’s difficult on an individual level. “The first step is recognizing impact, and recognizing that humans are imperfect evaluation machines,” says Hoffman. “That’s why we build in multiple rounds of feedback and multiple calibration points so that you can average things out and reduce as much bias as possible. One person can totally get a measurement of someone else’s performance wrong. Five people? Much less likely. Five people over time? Much, much less likely. Especially if you’re really clear upfront what the outcomes look like.”
Today, DigitalOcean is at a bit of an inflection point. In the company’s first four years, they offered one product. “It’s very easy to rally everyone around that one product,” says Warden. “We now have multiple products, and the things that we value in engineering teams now are slightly different than they were those first few years.” He has spent the past year determining what’s going to be most impactful from engineering teams in the future, and how to develop and reinforce those values.
Communication and collaboration: Software and products are usually built by teams, not individuals, creating a web of dependencies. The ability to thrive in a team setting is crucial. “The success or failure of any particular thing is codependent on lots of people, and so you can’t strictly evaluate one’s performance based on ‘hey, was that a great product or not,’” says Warden.
Prioritization: When you have one product, it’s very easy to have clarity on priorities. But with a portfolio of products and highly distributed team, DigitalOcean’s engineers and managers must be able to identify the largest opportunities to contribute value.
Breadth of experience: “Something we value is multiple experiences,” says Warden. “If you’ve only been successful, at some point I’m going to say ‘look, you haven’t screwed up enough things in your life.’ How are they going to handle a project that doesn’t go well? Shut down or be positive in overcoming difficulty?”
Focus on value, not production
What Warden values most in engineers and managers can be simplified down to their ability to have an impact. His goal is to unlock his team’s innovative spirit and get people thinking about what they are doing, why their work is important, and how they can maximize contributions - whether that’s by creating fewer bugs, solving harder problems, influencing more people, or raising the tide some other way.
“DigitalOcean is an incredibly user-focused business. If you don’t have an understanding around why what you’re doing is important and how to tie it back to a company goal or customer outcome that we’re seeking, then innovation isn’t happening,” says Warden. “As soon as you start always knowing why what you’re doing is important, you start having a better instinct about what’s actually important for customers and the business.”
Given the broad and potentially fluid interpretations of valuable, Hoffman recommends equally emphasizing behavior and performance. “A good development philosophy will be sure people understand upfront that we’re evaluating you not just on what you do but how you do it,” he says. “It’s the only way to create the space to empower top talent to perform in a rapidly changing marketplace. And as you get more senior, that stuff actually matters even more.”
This is impossible without a culture that empowers employees, offers intellectual and psychological safety, and encourages challenging the status quo. These ideas are core to DigitalOcean’s culture and Warden and Hoffman’s strategic visions.
“I’m very specific in that I expect the best ideas,” says Warden. “The things that need to get done come from all different places, so we try to create a culture that seeks those things out. And the people doing the work often have the greatest insight on things that we as leaders aren’t going to see.”
Leaders should avoid planning engineering schedules so tight there’s no leeway or room for innovation. “The last thing you want to do is say ‘you need to spend X many hours on this, Y on that,’” says Hoffman. Instead, set clear objectives, outline what good work looks like, and give employees breathing room. More often than not by setting the bar high and encouraging creativity to hit those goals, great employees will surprise you by the level of performance they achieve.
Personalize development and evaluation
Every employee likes to know exactly what they need to do to get promoted. Unfortunately, an equitable and effective team won’t have a checklist. This is especially true at collaborative, high-growth, fast-moving organizations.
“It feels like having some sort of formal list of boxes to tick, would be fair and remove bias, but it actually doesn’t,” says Hoffman. “There are so many factors that go into people being successful that if you just did that, it would, in fact just move the bias to evaluation and, be less fair.”
Warden adds, “It’s about what experiences have you’ve had, and whether you’re increasing your impact or value in some way. That’s unsatisfying the earlier on in your career, the fact that I can’t tell you what to do.”
Hoffman and Warden are currently working to move away from capabilities-focused evaluations, using a balance of qualitative and quantitative data points to gain deeper insights into each employee’s contributions and development; and using experience maps to help employees understand what successful outcomes they should be striving for.
Avoid metrics for metrics’ sake: Many employees and companies want performance to be metric driven because they think it’s fair, but Hoffman believes an over-reliance on metrics does the opposite. “You absolutely want objective measures, but making up numbers just for the sake of it is just moving bias from the evaluation process to the objective process,” he says. “You want objective metrics, and you need more of them as a company scales, but sometimes those are qualitative or qualitative. It can’t only be about the metrics but the connections between the numbers. Every input and output should simply be data points in a larger evaluation process. It’s exactly why humans are managed by people, not algorithms.”
Focus on enhancing strengths, not fixing weaknesses: Hoffman believes the most equitable and effective performance management framework (and the one that gets the most out of employees) is not one size fits all.
“You want to meet people where they’re at and leverage their strengths,” he says. “So if you’re good at one thing and she’s good at another, you actually don’t want to evaluate them against the same criteria. That may seem like super unfair, but that’s not the case. It’s actually fairer to find roles that you can help play to people’s strengths. An unfair process is ‘I’m going to set the same equally challenging expectations for you and this person even though you may be really at this and she may be really good at something else.’ Leveraging strengths is an important way to build up the needed diversity of skills and experiences across teams.”
Separate personality from performance: Critical feedback can often be biased by a manager’s feelings about their Engineers’ personalities. Managers, “should evaluate people in the context of a specific thing instead of evaluating the person,” says Warden. As Hoffman discussed, evaluate technical skill, value added, and personality/team fit separately, and, using 360° feedback, measure each as one of your many data points to evaluate performance
Emphasize experiences: Hoffman and Warden also emphasize what experiences engineers need to have that will be most beneficial to their long-term development. Growth plans and performance development for more junior people are more straightforward, but as people get more senior, it can get more difficult.
“You might have an engineer that was very successful on one product line,” says Warden. “You as a manager might want to keep that person on that team because they’re the best person to do the work that team has. But at some point that might become not the right thing to do. You might decide to put that person on another team, so they get more experiences and help them become even more impactful. You might take a hit on productivity on this team, or people will step up, but you really have to look out for how to give people a breadth of experience.”