Startups, Software Development and the Art of Bicycle Maintenance

Startups, Software Development and the Art of Bicycle Maintenance

LaunchDarkly CEO and Co-founder Edith Harbaugh biked solo across the U.S. in the summer of 2007. A decade later, her lessons from the road — from ways to grind through long stretches to how to approach forks in the road — still inform how she leads teams and builds software.

Break rides into legs.

When Harbaugh decided to try her cross-country bike ride, she was hesitant. “I had read a lot of blogs and didn’t want to have the humiliation of getting two weeks into it, finding it miserable, quitting and then having it be a failure,” says Harbaugh. “But I knew I really wanted to see Glacier National Park. So I titled my riding journal ‘To The Glaciers.’ As soon as did that my mindset shifted. I thought, ‘Okay, that's about a two-week trip, that's a doable and worth it.’ Once I got to Glacier National Park, I paused to ask: ‘Am I still having fun?’ Ok, I’ll go another week across Montana.”

Harbaugh broke that week (and all those that followed) into days and each day into checkpoints: a gas station for milk for breakfast, a little diner for lunch and a Subway for dinner. “Each pit stop was about 40 miles apart. That allowed me to lock in my pace and arrival times. I knew I was on a cross-country trip, but only experienced the glorious and gritty of the ride in 40-mile segments,” she says. “When you build software, deconstruct a feature in the same way. Break it into smaller steps — or minimum components that help you keep your pace and schedule.”

This concept isn’t foreign to software teams, but what may be is what each component and checkpoint should entail. “I think there's a belief that software is just the raw code. It’s not! It’s everything that connects to it: the documentation, marketing collateral and more,” says Harbaugh. “Focus on just the raw code is the equivalent of just paying attention to my bike, like checking the tires at a pit stop or adjusting my mirrors as I ride. If I don’t pay attention to the overall health of my system: my fatigue, signalling to cars at intersections or reading the weather, I’m in trouble. This isn’t just a metaphor for running a company, but building software. It’s all very connected. If you build a feature and no one knows about it, that feature doesn’t exist.”

Software isn’t just code but all that touches the product. Just like each leg of a ride entails the bike and all it moves through.

Harbaugh isn’t advocating that everyone on the software team must code, market the product, sell it — do it all from soup to nuts. But they must consider it as the build to be effective. “The bottom line is that nothing pains an engineer more than to have worked hard on something that nobody uses."

Every ride starts with zero.

An integral part of Harbaugh’s philosophy — as cyclist, CEO and seasoned product owner — is that every ride starts with zero. She had an odometer on her bike, and every day, she’d reset it to zero. It helped her:

With cycling — as with product development — you win when you operate with focus, but you want to win when you sense progress.

At the beginning of LaunchDarkly, Harbaugh championed weekly sprints with daily stand-ups that includes not just engineering, but other functions like marketing and sales. “Each day, we talk about what we did yesterday. Then, we discuss what we're going to do today. It keeps a one- or two-year project from feeling daunting. I also find that it creates a great deal of pride and accountability,” says Harbaugh. "If you haven’t found your cadence for sprints at a startup, don’t fix the sprint durations in stone off the bat. I spoke with Kevin Henrikson, an engineering leader who established two-week sprints, but soon changed them to be one week long. He noticed that all the work got done in the last 3 or 4 days of the two-week sprints.”

Write it down. You can always alter your route.

Harbaugh kept an online journal as she crossed the country by bike. “I knew this trip was something special, so of course I wanted to record what happened. I’d take photos and write entries throughout my months on the road,” says Harbaugh. “However, I realize now that I was also doing a daily stand-up — just solo. I was obsessively introspective about what was working and what wasn’t with my rides. I noticed quickly the difference between being tired and cranky, the latter which meant — nearly always — that I needed to eat. These are thoughts that many let pass, but when you write them down, you get to realizations and changes faster.”

At LaunchDarkly, each team has its task list that it uses as much as a to-do list as a capacity gauge. “As I said, we do a weekly sprint where we establish what we’re going to do — but before that, we review what we said we’d do. I try to set the tone that it’s fine if we don’t hit every mark, as long as we dig into why,” says Harbaugh. "The important aspect is writing it down. The mind just processes it differently when it’s written and team members can return to it asynchronously and as frequently as they want. Plus as a team, we can get a sense of if we are being too optimistic or not ambitious enough. Then we can adjust within a week.”

When new employees start at LaunchDarkly, Harbaugh tells them: at the end of your first week, you must publish a post on the company blog. “Try it out at your company. I tell them that they can write anything they want and we’ll edit it for grammar and spelling, but nothing else. It’s unfiltered and it’s fascinating,” she says. “Our ulterior motive is that we want to get a pulse of how people feel, how we’re doing and what they think of the culture. The public component is a big part of our content marketing strategy. Everybody in our company blogs — product, sales, ops, everyone. I want them to get in the rhythm of it. I ask the engineers to write an article once a month. It helps crystallize what they’re learning and is great marketing.”

You can't always A/B test the way forward.

From an early age, Harbaugh always liked riding bikes. As a teenager, she’d venture out 10 miles to the Washington Beltway near her home in Virginia. Soon it was 18 miles out to Reston, 24 miles out to Ashford and 30 miles to Leesburg, trying to get to the end of the 45 mile trail. “I wasn’t sure about 90 miles, so I asked my parents to drop me off at the end — 45 miles away — and biked one way back in,” says Harbaugh. “By 16 years old, I built up enough confidence and endurance to complete the entire 90-mile rides.”

The point is that it’s easier to test in some scenarios, than others. “If your primary focus is learning, test away. Think creatively as you are breaking projects into smaller tasks or organizing an experiment,” says Harbaugh.

But it’s different if you have a far-off destination you need to hit. On a cross-country ride — as in startup life — there are always diverging roads. “It was all forks — all forks. On any given day, there were so many decisions on the path forward. In some cases, like Montana, there was one road forward, but it was the exception, not the rule,” says Harbaugh. “In every state, I’d buy a paper map and scope out the best route. I wanted to avoid deserted roads and busy roads — both for reasons of safety. I looked for a road with a nice, wide shoulder and followed it. Sometimes that shoulder would peter out, and I’d have to make multiple course corrections.”

While Harbaugh could A/B test some parts of her journey — like which hotels gave her the best night’s sleep or gas stations had the best snack selection — she couldn’t with others. “If I were to hit my daily pace, each checkpoint for a meal and get to lodging by dusk, I couldn’t afford to double-back to take another road,” she says. “I did my best in the planning stage when I entered a state and pushed ahead. The key was to trust in your pattern recognition and then roll with the punches.”

Over Harbaugh’s career, this phenomenon has been most true with hiring. “You can't A/B test team composition. It’s not a helpful process to rerun a scenario of hiring a person. You can't. . That’s especially true with small teams at startups,” she says. "It’s about sample size. If you’re fine-tuning email marketing, you can do some complex A/B testing if you’re sending out 10,000 emails. You’ll never be able to hire — or interview — 10,000 people in a job cycle. I’ll paraphrase

You’re self-propelled and self-supported.

Ask any founder or sales lead or engineering head who’s accountable for building a company, customer base or app and the good ones point to themselves. The same goes for cycling. You’re your engine. It’s your two legs in rotation, propelling the bike and your body forward. That’s obvious, but here’s what Harbaugh learned from the road that helped her get better at that self-fueled forward motion: