In the late 1990s, when Kent Beck included “40 hour work week” in the original 13 practices of Extreme Programming, it was in the context of an industry that co-opted the term “death march”. Software development practices at the time were in a bad state: surprises around schedule were common (and almost always negative) because of deferred integration, poor code quality, unrealistic approaches to requirements, and a (false) belief that doing all the planning and design upfront was both necessary and sufficient.
The original idea behind Beck’s practice was to plan to work only 40 hours per week, and if you had to work overtime, be sure not to do so two weeks in a row. The practice came to be understood by some as “never work more than 40 hours”. It was eventually re-named “sustainable pace” as a more nuanced understanding of this important practice became common.
Today, the target of a 40-hour work week represents a progressive standard compared to the status quo in the United States. According to a Gallup poll, the average work week in the US is about 47 hours. I suspect the average for software development is higher than the Gallup poll found. A recent article in the Wall Street Journal describes the 40-hour work week as a radical idea for improving employee productivity and a recruiting advantage.
Benefits of sustainable pace
Sustainable pace has been a key part of our culture from the very beginning of the company. We see many benefits:
- It enables our makers to work productively and creatively for our clients. To no one’s surprise, studies show that working a lot of extra hours has a mean bite from the law of diminishing returns.
- While I believe work is a vital part of leading a fulfilled life, it’s not the only part. We work a sustainable pace so that we have plenty of time to be good spouses, parents, volunteers, and friends, and be happier people as a result.
- People who don’t work a sustainable pace are grumpy, un-collaborative, and no fun.
- Software creation is a profession and a craft. Our disciplines take years to master and constant work to grow and stay relevant. Working a sustainable pace leaves plenty of time for the professional development we all need to be doing.
- A sustainable pace means you’re prepared for those times when you have to work extra for a client or the company; it ensures you have those reserves.
A measure of success
Sustainable pace is one of the rewards that comes from having a predictable process. Predictability comes from skillful makers who:
- know what to build,
- manage code complexity,
- do good task decomposition and tracking,
- test all the way through a project,
- integrate continuously,
- validate with end users,
- communicate effectively with clients, and
- are disciplined about the meaning of “done”.
You can look at the ability to achieve sustainable pace as an indicator of successful projects. To the contrary, trying to fix a bad process, poor practices, or lack of skill and discipline by consistently working lots of extra hours is neither effective nor admirable.
Show me the data
Atomic has very accurate data for the activities of everyone in the company. We’ve always paid everyone on an hourly basis, and we invoice our clients by the hour so we can track against a project budget. Since our time tracking tool generates both invoices and payroll, we’re really disciplined about keeping track of our time.
Average work week
The graph below shows the average hours worked per week by each Atom for the last seven years. For each year (colored line), individual weekly averages are sorted from least to most and plotted on the chart. Each small circle is one person’s yearly average work week. The lines for each year have been stretched to cover the same horizontal distance, even though the number of full-time employees varies substantially from 2009 (15) to 2014 (35).
This unconventional use of a line graph helps visualize several things I found interesting and which are described below.
It’s interesting to note that the company average went up nearly 6% from 2009 to 2014. I’m happy to see that leveling off. On the low end of the scale, it’s good to see that the number of people not working full-time (2080 hours/year) has decreased over time. Clear communication around that basic expectation probably accounts for some of the increase in the average since 2009.
The high end of the scale shows a pattern I’ve observed over time: as your seniority, responsibilities, and past client relationships grow, you tend to work more hours. I’m happy that the upper end of our scale is still in the upper 40s, though that’s something we’ve learned we need to watch closely. We’ve recognized maintenance of prior projects as a source of stress for senior developers, and a challenge to our current structure.
In fifteen years, I’ve never observed our hourly pay motivating people to work beyond sustainable pace. And I’m glad that those putting in extra hours are compensated for that time. With the more common salary approach I suspect the 20% difference between the low end and the high end of average work week exists, but isn’t directly or fairly compensated.
Distribution of work week hours
Averages can hide a lot. Knowing that the 2015 weekly average across the company was 42.7 hours is nice, but doesn’t tell me how often people are working much more than that. The histogram below shows the distribution of weekly working hours for all full-time Atoms in 2015.
The distribution of weekly hours reflects both the week-by-week flexibility we offer employees, and the variability of client and company needs. In return for flexibility, we ask Atoms to put in extra time when a client or the company needs it. Being able to accommodate those needs is one reason being diligent about sustainable pace is important.
While the number of hours per week that is sustainable varies by individuals, this data is useful as a possible indicator of structural problems that can lead to burnout. Employees with the best of intentions and strong sense of responsibility may be trying to out-work what amounts to a structural challenge that consistently overwhelms their work day. It concerns me that 9% of the weeks worked last year exceeded 50 hours.
The advantages of maintaining a sustainable pace of work for all employees are clear. In the long-run, clients win, families win, employees win, and the company wins. Measuring, rather than guessing, and detecting potential structure problems, is one of the many valuable insights that accurate tracking of time provides my company.
I’m the founder of a company that I hope thrives well beyond my lifetime. Part of my job these days, fifteen years from our creation, is to pay close attention to company culture: what it is, how we communicate it, how we transmit it, and how we might usefully adapt it. When I recognized the need for some maintenance of our value mantras, I took it to the people who own them. I was both surprised and happy with what happened.
What’s a value mantra?
We use value mantras as a short-hand notation for what we care about and expect from each other and the company. Those things are a big part of our culture. Until recently we had five value mantras:
- Give a shit
- Teach and learn
- Own it
- Share the pain
- Act transparently
From the original four (“Act transparently” was added in 2012), our values are aspects of our culture that I observed and named. They existed many years before I named them. For example, I didn’t decide one day that I wanted a company where people cared about each other and their work, and where they valued learning, and thus told everyone to “give a shit” and “teach and learn.” I just noticed that these things were true and codified them in our mantras. We use our values frequently; having short, pithy names for them is quite handy. Read more »
We’ve been working for several years to improve the gender balance at Atomic Object, largely because we held “Diversity is good” to be self-evident. But recently I decided it was important to think through this belief and better understand it. Here’s why I think diversity is valuable, and makes my company stronger.
Diversity Means Liberty
Rather than an unquestioned end unto itself, I find it useful to think about diversity as an indicator of something valuable. Diversity in a profession is a measure of society’s success in achieving liberty. It indicates opportunity that can be capitalized on by anyone, regardless of gender, race, country of origin, age, etc.
Presuming there is no profession that is uninteresting to or unachievable by any particular group of people, then a diverse workforce is an indicator that there are no artificial barriers in place in the profession. The “natural” diversity of any given profession, absent of obstacles, is debatable. I believe all professions are attainable and of at least some interest to all groups of people. Increased liberty creates flexibility, greater productivity, and less waste from untapped talent. I think about this as labor market optimization.
In a liberal (in the British sense) society we place a value on an individual’s ability to make their own choices. This concept is famously noted in our Declaration of Independence as “Life, Liberty and the pursuit of Happiness”. Given how much of our lives we spend working, it’s hard to imagine achieving this goal with artificial constraints on career choice. A profession with very little diversity is a likely indicator of limits on liberty. Read more »
Atomic Object joined the ACLU’s Michigan Competitive Workforce Coalition following a couple of eye-opening experiences I had in 2015. The MCWC is organized around the goal of updating Michigan’s Elliott-Larsen Civil Rights Act to include sexual orientation and gender identity. What I learned was that on this issue, it’s not enough to be a company that doesn’t discriminate, and which welcomes and treats everyone respectfully. We’ve been all those things from our founding in 2001. What we haven’t done well in the past is to be explicit and clear about our values and behaviors.
The Atom We Nearly Never Knew
Months after a talented young developer joined Atomic, they shared the fact that as a candidate looking in from the outside, they weren’t at all sure they could be themselves as an employee and not need to hide their sexual orientation. They contrasted this very positively with how things turned out once they were actually an Atom. The fact that they were comfortable as an employee, and didn’t have to waste time and energy masking who they were as a person, isn’t at all surprising to me. The eye-opener was that they couldn’t be sure about this from the outside looking in, even after having gone through our extensive interview process and having read our diversity statement.
In a world of talent scarcity—certainly the one software design and development companies live in currently—it would be foolish to not accommodate a proven employee’s request to move to a part-time schedule if the job allows it.
I think the value equation for part-time knowledge workers is usually favorable toward the company. Presuming part-time work is desired by an employee, this should be a win-win situation. Read more »
I recently avoided a mistake I’ve made plenty of times in the past. The mistake I could easily have made was in how I reacted to an internal screwup. I attribute my better reaction to an experience I’d had a few hours before learning of the mistake we made.
Our marketing manager came up to me, a little upset, and told me we’d had an article published on one of our projects that inaccurately reported the membership of the team, excluding not only all of the designers, but both women who’d worked on the project. She’d heard from one of the designers who was justifiably irritated. Our marketing manager admitted she’d been busy, had handed the reporter off to a developer on the project, then moved on to other stuff she was trying to finish. Read more »
UPDATE 4/4/15 You can find the raw data I’ve used here at the bottom of the post.
I made an interesting discovery recently while spelunking in our time tracking system. The distribution of sick time, over all the people we’ve ever employed, very closely follows a logarithmic distribution.
All Atomic employees track their time on all activities, whether billable client projects or internal work. We’ve been doing that since we started, so our time tracking tool has nearly 13 years worth of data (679,263.25 hours, over 86 people). We find this data invaluable for analyzing our business, making staffing decisions, estimating, planning, and juggling responsibilities.
The graph below shows total sick time (blue bars) for all past and current employees, sorted by largest to smallest, left to right. The smooth blue curve is a logarithmic function. It has an R2 of 0.989 with the data. The red bars show absolute variance of each blue bar, versus the log function, at each point.
Atomic Object launched a new website back in November. While I believe the end result is very good, the time and effort it required surprised me. In the process I think I figured out why it is difficult to market companies like Atomic, and why it’s often done poorly.
I’ve come up with a HierArchy of Relative Difficulty (HARD). Where your company exists on the HARD scale determines how much time, money and expertise you’ll need to effectively market, or conversely, how likely you are to be marketing your company poorly.
No slight on the widget makers of the world, but their marketing challenge is pretty straightforward. The picture below illustrates the situation. Companies that make a concrete product have a straightforward marketing challenge. Describing the product isn’t difficult. It has known features, materials, quality, costs, and benefits of use. The marketing can be directed to the person who will both buy and use the product.
A service is inherently more abstract than a product. Service companies therefore have a tougher job in telling their story. The features and benefits of their service are less concrete. Pricing may be more complicated, and the relationship with the customer is more intimate and less transactional. Quality is more subjective. There are no materials.
I’ve always believed that innovation comes from people who care. When you care about your profession, your clients, your peers, and your company, you’re never entirely happy with the status quo. You ask questions like, Can this be done better? Quicker? Cheaper? Is there a different product that would help or delight people? The self-applied pressure to improve, combined with creativity, brains and experimentation, drives innovation.
Transparency is a cultural attribute of a company that, when exhibited by managers and leaders, engenders trust in employees. Trust is a critical element to nurture caring people. After all, caring makes you vulnerable, and if you don’t trust the company or the people you work with, your caring is likely to wither, or you’re likely to move on.
Transparent environments have some downsides. I’ve previously written about their ability to spread anxiety, sow confusion, and slow decisions down. These negative outcomes of transparency reflect my viewpoint as the leader of Atomic Object. Research reported in the Harvard Business Review this month helped me see some negative aspects of transparency from an employee’s perspective.
In the Transparency Trap, Ethan Bernstein, an assistant professor at Harvard Business School, describes a simple, powerful model for understanding the downsides of a transparent environment. Bernstein observed four types of boundaries in companies that effectively balance transparency and privacy:
- Boundaries around teams
- Boundaries between feedback and evaluation
- Boundaries between decision rights and improvement rights
- Boundaries around time
A company can’t reach the milestone of celebrating its 100-year anniversary unless it can out-live its founder. In fact, it can’t even get to the lesser goal of outliving the founder’s retirement unless it pays attention to succession planning. This blog post is about the path Atomic took to reach this point.
A Multiyear Effort
Handing over responsibility and daily operations for our first and largest office was a multiyear project. We’ve now achieved the significant milestone of replacing me as founder in the managing partner role for Grand Rapids.
To be clear, I’m giving these phases names, durations, and significance retrospectively; the only “plan” I made was to achieve the goal, and the initial impetus to start was feeling overwhelmed.
In 2008 we were seven years old with about 20 Atoms in the molecule. I was selling all the work we did, planning capacity, actively involved in billable work (35% of my time that year), managing people, networking, marketing, wrestling with issues around growth, and working on broadening ownership. I got some help on some of these things, but what was on my shoulders was more than a full-time job (2,370 hours that year), more than I could do well, and pretty stressful. Read more »