People are the only valuable asset of an innovation services company. While reputation, client list, culture, standards and tribal knowledge are also valuable, those all derive from and are maintained by people. Considering how important people are to Atomic Object, it seems crazy, when I think about it, that I don’t have a reliable way of measuring the state of our people. I’ve put more thought and effort into financial and technical measurements than people measurements.
We measure things like test coverage, project test status, billable hours and utilization, and we track revenue and profit margin. These are all strongly determined by or related to successful projects and happy clients. And happy clients, in turn, presuming you employ people with the right skills and expertise, derive from content, satisfied, happy employees. This might be the most important thing I should be measuring.
So what should I try to measure in people? Contentment? Satisfaction? Fulfillment? Joy? Happiness? — since I’m not sure, I decided to stick with the most general term, which is happiness.
Admittedly, measuring happiness of a group of people isn’t straightforward. But suppose I could reliably measure the aggregate happiness of my company. What would it tell me? Would it be reliable? Would it be helpful?
How I’d use a happiness metric
If I had a reliable measure, over time, of the aggregate happiness for the company, I believe I’d have a valuable tool to help make and time the implementation of big decisions, and a measure of the impact of significant policies and new initiatives.
While the absolute value of my happiness metric would be meaningless, I believe changes of that value over time would be meaningful. Since the metric I’m imagining is an aggregate metric, I would rely on first-hand investigation, conversation, and group discussion to understand the root cause behind the change I was measuring.
Part of the job of a company’s leaders is to keep an eye on exogenous forces (clients, the market, technology, the world in general) that may impact the company’s health, or which present a valuable opportunity. These pressures or opportunities from outside the company sometimes require change inside the company. Given that change is generally stressful for people, and the absorption rate for change by any group of people is finite, tracking happiness over time could be very helpful when trying to decide when the introduction of change will be successful or particularly stressful.
Strategic initiatives such as new service offerings, major new internal policies, expansion to new markets, or new partnerships, put stress on the company. They may require extra work, or establishing new relationships, or some travel burden. Tracking happiness over time would give me a measure of that stress, when it peaked, and when it receded.
I’d like to have a happiness metric to gauge the long-term impact of structural changes like company ownership, changes to our service offering, the kind of clients and projects we take on, or how we schedule and assign projects.
How I do this now
I measure happiness of the company now in a very ad-hoc fashion. I watch how people interact after standup. I observe laughter and frowns. I chat casually with probably half the people in our office in any given week. I ask others how they think the company is feeling. I watch how well attended company parties are. I visit with spouses. While these measures are helpful, they are inevitably distorted by sampling error and my own feelings.
I could be disciplined about talking to everyone in the company regularly, in a formal fashion. But in a company with 40 or so employees, that would not only take a significant amount of my work week, but it would still be subject to my personal interpretation, relationships, and position. It would be difficult to quantify consistently, and it might be hard to learn the truth, given my role.
A better measure?
What if instead I anonymously asked every employee a simple question?
How happy do you feel right now?
I imagine an interface that makes it possible to answer this question on a simple three or maybe five point scale in less than one second.
I don’t want people to spend a lot of time thinking, I want them to simply react, as close to unconsciously as possible.
I don’t want to ask people to segregate work from personal happiness.
I don’t want people trying to average out their feelings over a period of time (e.g. month, quarter or year).
I want to make it easy to answer spontaneously.
I don’t want people to have to write in order to respond.
I will rely on the aggregation across people and time to make the metric meaningful.
I’m envisioning something that integrates with our time tracking tool, PunchIt. Since everyone in the company interacts with this tool multiple times per day, it seems like an obvious place from which to sample the population.
The image below is my crude mockup for how this might look superimposed on PunchIt. A single mouse click on one of the faces would send the result anonymously, to a web service, dismiss the happiness poll widget, and return focus to PunchIt.
My guess is that sampling everybody once or twice a week would be sufficient to capture meaningful data while not making the polling a burden. I believe that varying the time when the polling occurs would both improve the accuracy of the data as well as make it less obtrusive.
An immediately obvious measure of happiness is retention. But waiting until you lose people seems dumb. Presumably a significant, sustained, drop in happiness would indicate that we’re at risk of losing talented people. The aggregate measure I’m contemplating wouldn’t tell me who I’m at risk of losing, but it would be a signal that something was up, and that I should investigate. Atomic doesn’t have a retention problem*, but we don’t have a profitability problem either, and I measure that.
In my casual research on what other people have done to measure employee happiness, I’ve found companies using net promoter scores, traditional surveys (e.g. 65 statements, 2 open-ended questions), and short interviews or reports. I’ve not found anything as simple as a single question with a spontaneous response.
Displaying the metric
I’m envisioning something on our open office information radiator that charts aggregate happiness over time. Displaying this where everyone can see it would help all of us understand the mood of the office, and know when we should be having conversations about what’s going on.
If the aggregate measurement was accurate, I’d also have a way of gauging the significance or breadth of an individual’s observations about the broader group. I’d know how much credence and weight to give to statements like “the company is stressed”, or “everybody feels this way”, or “I hear a lot of people grumbling.”
I might see interesting patterns. For example, I might learn that I can time company meetings or important emails (that that have the potential to trigger stress or confusion) for parts of the week where aggregate happiness is higher.
Consider the two scenarios below. In the graph on the left, the mood of the company has improved over the long-term average (dashed line). That’s not only nice to know, but might tell me that the change introduced recently has had a positive impact. It might also tell me that the company is mentally able to tackle a new initiative or wrestle with a difficult challenge. In the scenario on the right, the aggregate happiness has decreased markedly and is trending down. I’d be arranging company-wide conversations to find out what’s bothering us, and whether we should be doing something about it. I’d be worried about stress levels and looking for ways to smooth things out and not to introduce new stresses.
What I’m proposing is a radically simpler measurement of overall company happiness. It would be a metric that is subject to all sorts of factors (health, traffic, mood, caffeine, personal relationships, clients, bugs, sunshine, colleague interactions, etc); a measure of the whole person. My hope is that it would becomes meaningful and useful as these tiny, easy-to-do, spontaneous samples of personal happiness are aggregated across many individuals over a period of time.
I don’t intend to stop gathering anecdotal evidence about the company’s overall happiness. And of course no metric replaces the action of conversations with individuals, groups and the company as a whole. I see my happiness metric as the canary in the coal mine — an early warning about something dangerous in the air.
Am I nuts? Do you think tracking this metric over time would provide me meaningful data to gauge the overall emotional state of the company?
*In the last five years we’ve had some involuntary turnover, but we’ve only had two people who left because they thought they’d be happier somewhere else.