Last week was my blogging debut, and wow, what a reaction. However, I think the most interesting discussion actually occurred in the comments to the previous blog post, The Hunt for Engineers: Expensify and the surprisingly difficult challenge of finding the right people. I think the controversial statement in the previous blog was that we look for people with “a great work ethic, almost to the point that some of your friends might call you masochistic”. But what does that really mean?
Some of the commentators thought that it was some sort of codeword for working people to death. And when I mentioned we don’t discriminate by nationality (we’ll hire you regardless of visa status), the assumption was it meant we hire people who are desperate and kept in check by fear of deportation. In other words, most commentators assumed the worst interpretation of every single word because, surely, who in their right mind would work so hard as to bewilder their friends?
The answer is, obviously: people who work harder than their friends by choice. People who have so much passion and ambition that they can’t be stopped from working, and who pursue that ambition with such vigor that their friends simply cannot understand.
But all that is relative. How you interpret those statements depends on a lot of assumptions, and different assumptions lead to wildly different conclusions.
For example, some people claim startups work 70 hours/wk in an endless period of “crunch time”. Others suggest that nobody can possibly work more than 17 productive hours/wk. There are so few hard numbers to analyze that everybody works from anecdote, myth, and legend — leading to wild accusations without substantiation. So in an effort to move the discussion forward, let me bring some data to light. Behold, for the first time, the Expensify workweek:
As for how to read the chart, the X axis is weeks (since Expensify’s start), and the Y axis is hours (number of “real hours”/wk). A “real hour” is “an hour spent doing productive work” — graded by the honor system. Like any real-world dataset, there’s something for everyone, and I’m sure you can argue any conclusion you like. But the conclusion I’d like to argue is, on average, we as a team work 50 hours/wk, week after week. Often people work more, and very often people work less. But nobody has ever worked more than 70 hours/wk, and nobody puts in substantially over 50 hours for an extended period of time. This is what hard work means at Expensify, and I’m very curious to hear how it compares to your experience.
I’m sure we’re not alone, but Expensify is the only startup I know that tracks hours. (If you do, please share your data.) Unless you’re actually billing hours to clients, there might seem little reason to. But tracking hours gives two, extremely powerful benefits:
- Everybody knows, objectively, when they’re not living up to the team’s expectations, and by how much.
- Everybody knows, objectively, when they’ve gone above and beyond their obligations, and by how much.
I have no doubt that most people will latch on to the first bullet as if it’s the only one listed. Furthermore, they’ll say it’s all a tool to convince people to work harder, putting in endlessly more hours toward an insatiable corporate appetite. But as hard as it might be to believe, the main motivation behind this system was actually the second bullet: to convince people to work less.
See, as we’ve made patently clear, we look for a very unusual sort of person. The sort who has done more before joining college than most people will ever do after. People who have audacious plans for their life. and who will stop at nothing to succeed. These are over-achievers by birth; people who go home after a hard day’s work and pick up some project that’s even harder. These are people who move mountains on a daily basis, just to see what’s underneath. You might not know any of them. But they exist, we have several, and we want more.
But a problem with this sort of person is they frankly don’t know when to stop. They’ll work themselves to death one week, and then burn out the next. They’ll casually do something in an hour that’d take an average person 10, but then spend 10 hours screwing with the CSS. Without the right sort of environment, this person will not only fail to excel, but will genuinely underperform. They’ll get bored and slack, be disruptive, or tend toward outright insubordination. I’m as guilty of that as the next.
So the goal is to channel this enormously powerful but unstable force in a positive direction. And half of that is giving clear guidance. But the other half is setting very clear expectations as to how much of their time the company expects, and genuinely encouraging them to take the rest off.
I don’t want someone to work 70 hours/wk. I work nearly as hard as anyone on our team, and I know it’s incredibly hard — nearly impossible. I don’t know that I’d believe anyone who said they put in 70 productive hours in a given week. I don’t want anyone to feel they need to attempt that impossible task, or to pretend they’ve accomplished it.
But at the same time, I do want everyone to work 50 hours/wk. I’ve generally maintained at least that for years, so I know it can be done. It’s not easy — it requires sacrifice — but not nearly as much as you’d think. I’m plowing through 6 seasons of Lost, I have dinner with my wife every night, I recently got married in Italy, I spend weeks overseas every year, and I walk my amazingly cute beagle on the beach most weekends. There are 168 hours in a week; if you can’t work 50, you’re doing something wrong.
In fact, the primary purpose for all this (I’ve been doing it for over 5 years now — it only takes seconds each day) was to create an objective tool to achieve a desired work/life balance, in large part to live up to my family obligations. I want to work extremely hard, but then go home with complete confidence that I’ve done right by my startup ambitions, and can therefore relax without guilt. Setting clear goals encourages me to work the right amount, day after day, without anybody looking over my shoulder — but then to stop and do something else.
In short, I want everyone to maintain an aggressive pace, one that seriously challenges their work ethic, but one that I know from personal experience can be maintained over the long run without giving up the good things in life. I want this because this is what it takes to compete in the startup world. It’s what it takes for us to collectively achieve our individual long-term goals. It’s what we need to do to succeed in this venture and lay a strong foundation for the next.
It’s not for everyone. I expect many people will be horrified to read this post. And for those who aren’t horrified, I bet your friends are. But that’s ok, because we don’t want to hire your friends. We want to hire you.
Expensify Blog 
If your company needs your programmers averaging 50 hour weeks, you’re doing something wrong.
If you’re really hiring the best of the best, you’re getting much more than the 25% boost you hope to get from some extra hours anyway. Joel Spolsky would say you’re getting about a 1000% boost just from hiring the right people: see http://www.joelonsoftware.com/articles/HighNotes.html.
These 50 hour weeks cut into personal project time, too, which is a big deal for a lot of programmers—especially the hyper-motivated types you’re wanting to hire. It’s especially toxic because you build expense reporting software, which is the kind of stuff in-house programmers for huge mega corps selling laundry detergent are forced to build. Either I’m missing something big in the world of expense reporting, or the people you want to hire are going to want to work on something else on their own time.
Maybe it’s not an issue for your team. If it’s not, either expense reporting is a lot more awesome than I realize, or the people you’re hiring aren’t as passionate as you think they are.
For another take on working long hours, read http://www.kenrockwell.com/business/two-hour-rule.htm. A working sense of humor is definitely required, though, as is the willingness to actually read the whole page.
On the other hand, surely you’re paying *at least* 25% more than your competitors, right? Maybe I should be willing to suck up the extra 10 hours a week and apply for a job instead.
Hi Jeremy, thanks for writing. Yes, we compensate very aggressively, especially on the equity side. As for expense reporting being more awesome than you’d expect, I guarantee that’s true: it’s a lot more awesome than *I* ever expected, too.
It’s a sort of long story, but we actually got into expense reporting precisely because it sounded boring. Our first product was an actual, honest-to-god payment card, and that sort of thing requires integration with a lot of banking partners. If you’ve ever dealt with banks (from an integration perspective) you’ll know that they’re hyper-adverse to risk. So how do you get around that? That’s right: you make yourself sound really, really boring. Hence expense reporting.
But what I didn’t realize (and I don’t begrudge you for a moment not realizing it either) is that expense management sets at the crossroads of the most amazingly juicy data, dollars, and user demographic you could possibly want. And because the financial space as a whole is so reluctant to innovate (because it refuses entry to anybody with a new idea due the potential for risk), there are amazing opportunities everywhere we turn — opportunities essentially nobody else is aware of because they’re very hard to see until you’re already here. Opportunities that require really great people to take advantage of.
Anyway, I don’t blame you for being skeptical. As I heard one person comment “Why does it take the best people to do something you can do with Excel?” That’s fine; I’m happy to fly under the radar for a bit longer.
But when you see what we’ve got in store, you’ll kick yourself. So I’d love to talk with you before then and see if it’s a good fit. Check out https://expensify.com/jobs — or even email me direct at dbarrett@expensify.com if you want to learn more before applying.
Thanks for reading, and I hope to hear from you soon!
Equity? From you guys? Another clown saying they’ve got the next billion dollar company.
Worthless – you want the hours, the investment, pay for it.
Nice one, Dave. You nailed it.
Keep up the blogging. I’ve always wondered how the ‘man behind the curtain’ of Expensify thinks, especially considering your quirky tech habits. Your product makes one of the most boring tasks exciting for me to use and recommend.
Very curious about what you’re using to track time. Some web app? Plain ‘ol Excel? And when you’re tracking time, do you categorize how you’re spending it? What’s most interesting about this chart is how your weeks are starting to get less varied, the band is getting tighter.
The problem, of course, is how to quantify productivity.
Agreed: “productivity” can’t be measured on an absolute scale. Two people might get vastly different amounts of work done in a single “productive hour” for any number of reasons (skill, experience, difficulty of the problem, how they define “productive”, etc) so they can’t be directly compared. But the point isn’t to measure “total productivity” of one person versus another — that’s impossible. Rather, the point is to give people a tool to evaluate their current productivity against *their own past* productivity, and then to simply ask that they use it.
Pashmina – It doesn’t need to be advanced. We just use a Google Spreadsheet with one column per person, one row per day. Then we have other sheets in the spreadsheet that show historical averages, calculating the difference between how many hours you’ve worked versus how many you did (if positive, that’s extra time to take off next vacation), etc.
It only takes a few seconds per day, but those seconds force you to “get real” about your productivity in a way that is easy to overlook if you’re not actively paying attention. It’s amazing how often I’ll log my hours and think “Seriously, that’s all I did today?” After a while of that you start to think “Maybe I shouldn’t be reading so many blogs (except for the Expensify blog, naturally)” and “Maybe I should start planning for some of these ‘unplanned distractions’ that seem surprisingly common”, etc. More than anything it’s just a daily reminder that “I have control over my productivity, I don’t need to just accept my unproductive nature as a foregone conclusion.”
You have a few problems with getting high quality engineers here. First a high equity compensation package suggests lower average salaries. But more importantly, the highest quality algorithmic coders like a deep challenge in terms of complexity and scale. Writing software for a cloud computing platform that operates on one million servers or more doing six billion transactions per second seems complex and interesting. The best coders have that as an option.
50-70 hours a week is not work life balance. if you expect 50hrs a week, your engineering process is terribly broken. And you know nothing about the quality of rest and a clear mind. People need family time. They need time to rest. And exercise. And form other hobbies. Well rested, well rounded engineers are what i hire. not 50 hr a week robots, passionate or not.
Based on the graph, you are in fact working roughly 45 hours a week. The standard deviation is such that you work over 50 hours / week roughly 25% of the time.
I did not process the entire graph. I took the 50 latest samples. The values from the first year may distort the results, especially since they are a product of less individuals.
Since you only hire squirrel-hunters coding all day in machine language eschewing the benefits of modern IDEs It’s no wonder that they have to spend 50 hours a week…
I’m sure your turnover costs are nice and low too? Are you even accounting for that? A good manager will make sure his team isn’t working much more that 40 hours a week because they understand the human cost of those hours. And they will have no surge capacity when they need it. ugh…
Geordie – It’s true that we don’t keep a lot of manpower in “reserve”. However, we also explicitly avoid the “boom/bust” cycle, and aim for a continuous amount of activity week after week.
Assuming you find Superman, why would he want to work for you?
We have a strict “no-kryptonite in the office” policy.