I made many mistakes as we grew Atomic Object. Since we turned ten years old a few months ago, obviously none of them were fatal. For example, not getting help in my job sooner was one of the bigger mistakes I made. Seeing many small mistakes as evidence of a smart strategy is something I’ve just recently come to appreciate.
Mistake is just another word for a failed experiment. The purpose of experiments is to confirm or deny a hypothesis, or in other words, to learn something. Saying that we made a lot of mistakes is also saying that we ran many experiments. For example, we built a web CMS product back in 2002. We invested time and effort to try an open compensation system. We invested in various strategic partnerships. We hired someone into a leadership position from outside the company. We created an isolated design team. We bought an ad in a trade magazine. We took over the maintenance team for a large customer. These are a few of our failed experiments. None of them worked the way we anticipated, but we learned an awful lot from these “mistakes”.
Lately, I’ve been working on a second round offering in what I call the Atomic Plan. My goal is broadening ownership in Atomic Object. The first round offering was made in May 2009. That sale of equity brought in seven new owners and started us down a very interesting path of succession planning, leadership and experiments with governance. It hasn’t all been smooth, but we’ve made some interesting discoveries and some significant progress.
As part of this second round, I decided to look again and see what the literature on employee ownership had to say. I had previously familiarized myself with the technical body of knowledge around employee stock ownership plans (ESOPs). My search this time has turned up more interesting sources. So far the best book on the subject that I’ve found is Equity: Why Employee Ownership is Good for Business by Corey Rosen, John Case, and Martin Staubus.
The profit and loss (P&L) or income statement is critical for understanding your business. Every quarter we close the books, analyze the P&L, compare against our economic model of the company, and prepare a report for the company meeting. We usually notice some anomalies. It can take quite a few hours digging around trying to explain why Cost of Sales varied unexpectedly, why metrics like revenue/employee varied, whether or not expenses were in line with expectations, etc. We improved this process considerably when we started adjusting for WIP and accrued wages.
We’ve made four more changes this year that have made our P&L more useful, and simplified our quarterly financial analysis and reporting process.
I’ve written about some of the difficulties of describing your “why” as an innovation services firm. I began thinking about our “why” last year when Melissa Bugai first introduced Simon Sinek’s talk to me. The process I went through at Atomic to get to a consensus description of our “why” made me realize something interesting. After explaining Simon’s idea, I asked various Atoms the question, “What do you think the ‘why’ of AO is?” I almost invariably got answers to an equally important, but very different question, namely, “Why do I work at AO?” I’m now calling these two questions the “existential why” and the “practical why”.
Simon Sinek posits that successful companies know “why” they exist, outside of “what” they do for customers, and “how” they do it. His TED talk seems to be everywhere recently. I’ve seen it on Detroit Venture Partner’s website, heard it in John Hwang’s presentation at BarCampGR, and saw it used in as a part of the Momentum program last summer. Melissa Bugai first brought Simon’s idea to my attention early in 2010.
I think having a strong, shared, “why” is a good indicator of a company’s potential for success. It’s one component of culture, the third and critical ingredient necessary to create a sustainable company. Your “why” is different from your “what” and “how”. Apple exists to challenge the status quo and simplify our lives — that’s both easy to understand and explain. Disney believes in good clean family fun — everyone in the world can appreciate it. Facebook helps people share things with their friends. These statements say nothing about “what” these successful companies do or “how” they do it.
Great Not Big is about software development companies that choose to pursue excellence over largeness. The kind of software development companies I’m interested in are examples of innovation services firms — they sell their expertise and capacity to co-create innovative products with and for their clients. While my blog’s name is pretty self-explanatory, I thought it might be interesting to share the story of how it came to be.
Read more »
I suppose it makes sense that people are the most challenging aspect of running an innovation services firm like Atomic Object. Even when you take away the obviously critical people we call clients, my greatest struggles, the peaks of my happiness, and the valleys of my despair, can mostly be attributed to the amazing people we employ. Operations, finances, strategy, sales, facilities, technical practices — these challenges are as nothing compared to leading and managing people.
My personal management style is consultative, collaborative, and participative. Talking through problems and opportunities with other people is a natural part of my decision making process. I feel like we make better decisions together than I would alone. Even when it comes down to a decision I alone make, I seek other’s perspectives and ideas beforehand. This style requires my employees to engage in complex issues, question me, take time to understand, and sometimes disagree with me. I want to make it clear, the upsides of working with people willing to seriously engage like this far outweigh the downsides, but at times I experience frustration along the way.
One of the original 12 practices of Extreme Programming is “40 hour work week”, later renamed “Sustainable pace”. Atomic’s stuck closely to this practice as a company from our very first days. In a nutshell, it says individuals, and hence teams, will be most productive if they work at a sustainable pace. Sustainable pace acknowledges the world is not perfect. If you need to work beyond your sustainable pace one week, you shouldn’t do it the following week. In the industry that coined the term “death march”, this was a big deal. While sustainable pace is clearly valuable to the people on your team or in your company, the underlying motivation is the interest of the customer’s project and budget. Unhappy, stressed-out, tired, grumpy people don’t do their best work. They make mistakes. They cut corners. In the end, an unsustainable pace costs the customer money and jeopardizes project schedules.
Until recently I’d never seen mention of research supporting the practice of sustainable pace. It turns out there have been a lot of studies done, some dating back to the early 20th century. A recent article in The Atlantic brought my attention to a summary of eight studies, and states some conclusions succinctly. For example,
Overtime binges lead to bursts of output that exert a hangover effect in later days. Study after study indicates that short bursts of attention punctuated with equally deliberate breaks are the surest way to harness our full capacity to be productive.
I found the graphs on productivity vs work hours interesting (the embedded slide deck is from Lost Garden). In particular, the shape of this curve appears to be the same across studies, which indicates to me that there’s an underlying explanation independent of the type of work or study done. That underlying commonality is of course the complex mix of physiology, psychology, and brain chemistry that defines each of us. I believe the results of these studies are even more relevant and important to innovation services firms because of the nature of the services they provide. The importance of creativity and the damage that poor quality can do strongly suggests that you want people working at their best all the time.
I’m quite proud of Atomic’s record on sustainable pace. A while ago I wrote up some detailed suggestions for potential clients on how to effectively use an RFP to select vendors for software development projects. One of the elements I suggest looking at is whether a firm respects sustainable pace. The research cited in the Lost Garden blog post seems to confirm the wisdom of this criterion. As part of writing my RFP post I pulled data from Atomic Object’s time tracking system. The chart below shows seven years of data. The data charted is a histogram of the number of hours our makers (designers and developers) work per week. Over the last 7 years, our makers worked 35-45 hours per week 74% of the time.
Distribution of hours worked per week shows Atomic makers stick closely to the practice of sustainable pace.
Sustainable pace isn’t the only way to explain Atomic’s success, of course, but I think it’s a major contributing factor to our excellent track record, high employee retention, and creativity. As I mentioned at the start, we believe strongly in sustainable pace. It’s nice to see the research, our data, and our rhetoric all in agreement.
Innovation services firms can use thought leadership marketing very effectively. At Atomic, one of our core value mantras is “teach and learn”. We practice this value and further our marketing efforts through publishing, presenting and blogging. Since we’re analytical, somewhat data-obsessed people, we of course attempt to track the effectiveness of our thought leadership marketing through Google Analytics (GA).
Combined, the Atomic Object website and blog receive in the neighborhood of 5000-6000 visits per month. The blue line in the graph below shows the visits for every day in August 2011. GA can tell you all sorts of interesting things about visitor behavior, which pages are popular, where visitors live, traffic from browsers, etc. You can spend a lot of time exploring and pondering the meaning of the data.
Unfortunately, GA can tell us almost nothing about the most important visitors to our website.
Total website visits and potential customer visits by day.
I estimate we’ll handle approximately 300 sales opportunities this year. Even assuming they all came through the website (they don’t, obviously), and further assuming that 10 times as many potential customers look at our website as actually contact us, the visits from potential customers is a drop in the bucket compared to the overall site traffic. The green line in the graph shows theoretical potential customer traffic modeled on the assumptions above (10 potential/actual * 300 actual/year * 1 year/12 months = 250 potential/month). This drop in the bucket is probably exaggerated, as I think it’s not terribly likely that our site sees 10 potential customers for every actual contact. Multiplying by 10 also covers the other very important “customer” of our website — potential new hires.
When the most important traffic to your website is tiny compared to the total traffic, it’s difficult to use GA to measure the effectiveness of the site for its ultimate purpose. For example, GA can tell us nothing about which browsers our potential customers are using. And the content I wrote 15+ years ago on OO aggregation vs inheritance probably isn’t too interesting to potential customers, even if it is one of our most visited web pages. Do potential customers see our client list? our bio pages? our case studies? GA is silent.
While there are some clear advantages to having a lot of traffic, I’ve found it very difficult to pull the “signal” of potential customers out of the “noise” of overall visits. It’s a good reminder that what Eric Ries refers to as vanity metrics applies more broadly than just to startups. In the end, if the phone doesn’t ring or the contact mailbox is empty, all the traffic in the world is for nought.
If you find yourself up for parole in Israel, you should pick the time of day you make your pitch very, very carefully. An article in The Economist back in the spring described a study of parole applicants and their chance of being granted parole as a function of the time of day the case is considered. The data clearly show you have a much lower likelihood of being paroled if the judge hearing your case is tired or hungry. The graph below (a simplified sketch as The Economist wanted $460 to use their graph) shows how incredibly strong this relationship is. The chance of being paroled decreases as the number of cases the judge has heard increases, and meal breaks reset the chance from being very low to quite high.
Chance of parole as a function of the number of cases considered over the course of a day
One of the questions addressed in the study was what explained the strong correlation between outcome and time of day. Was it the judge’s hunger or his fatigue that mattered more? The researchers found that the number of cases heard by a judge is more strongly correlated with denying parole than the judge’s blood sugar level. Making parole decisions is mentally taxing. The more decisions a judge has made, the more tired he becomes and the more likely he is to start looking for easy answers (in this case, to say “no” is easier, safer, and preserves future options). Making hard decisions, in other words, is fatiguing.
Read more »
