Developer Happiness: What Makes Developers Happy?

First, thanks to everyone who took the time to complete the Developer Happiness Survey originally posted at DZone. I can safely conclude from the >500 responses that the average respondent (as if any are average) is a moderately satisfied yet fairly paid to underpaid engineer who generally avoids user groups and extracurricular coding while boasting a manageable commute to a somewhat quiet office where mildly challenging development is performed by an average team (led by average management) using an ill-defined development methodology and unclear requirements employing subpar/mediocre tools practicing source control/continuous integration and consistently updated schedules, where testers test (and bugs are found and promptly fixed), interviewees don’t code, and software is a source of profit.

Does that sound awesome or what? Forget all that for now.

About The Survey

I designed the happiness survey based on two theories. My primary goal was to test the accuracy of nearly twenty years of anecdotal evidence, amassed over thousands of conversations with developers usually looking for new work (because I’m a recruiter, and that’s why people talk to me). When you’ve asked the questions “Why are you leaving?” and “What are you looking for?” thousands of times, you soon discover that there are only a handful of common answers cited.

Second, I was curious how well the individual elements of the Joel Test and a combination of the Joel Test questions might “stack” up (see what I did there?). I peppered the survey with variations on 8 of the 12 questions Joel Spolsky wrote back in 2000 to try and rate the quality of a software team.

9 of the 19 questions offered an “average” answer option, while the other 10 questions were binary. The question about overall job satisfaction (“happiness” – and yes, I know these aren’t exactly the same) offered an average option, which explains why % happy + % unhappy < 100%.

Survey Results Breakdown

Compensation

Just under half of you feel underpaid, and an almost identical percentage feel they earn at market rate. Both underpaid and at market respondents were mostly in the average job satisfaction category, but among the underpaid the unhappy outnumbered the happy 4:1.

Those who felt they were paid market rate were equally likely to claim happiness and unhappiness.

Only 2% of responses claimed to be both overpaid and unhappy.

Challenge

Many developers cite a lack of technical challenges as a reason for leaving their jobs. Half of you claim to be still learning at the workplace, and dissatisfaction among that “I’m still learning” group is a mere 11%. Half of the underchallenged are unhappy, while only 2% are happy when not learning on the job.

Tools and Stack

Only 25% reported that your employer uses the best tools regardless of price, and a combined three-quarters identify as using a rather standard (48%) or cutting-edge (26%) tech stack. Less than 1% of all respondents reported being happy while still using a dated stack. Only 12% of those who use the best tools are unhappy, compared to the 38% dissatisfaction rate among those using second-rate tools.

People

The competence of co-workers and management is often cited as important to job seekers, and the numbers here seem to validate that observation.

Regarding co-workers, three-quarters rated your team as average (45%) or above average (33%). Just under half of you also claimed to be the most knowledgable person on your team. 10% of devs on above average teams are unhappy. Compare that to 3% of those on poor teams being happy (more than half are unhappy), and the value of a good team is clear. Being the best on your team also comes with a 1/3 dissatisfaction rate, which may be felt by those who can no longer learn from peers.

As for management, about one-third of you described your bosses as ‘mostly incompetent or dysfunctional’, which came with a whopping two-thirds dissatisfaction rateLess than 1% of all respondents reported being happy under bad management or unhappy under competent management.

Cost vs Profit

I have heard developers regularly expressing more interest in working for companies that either build a software product or are at least a technology business, as opposed to firms where tech is a cost of doing business. The ratio of happy to unhappy developers at tech firms is unremarkable, but dissatisfied developers outnumber the satisfied by nearly 4:1 where software and tech is not the focus.

Remote Work and Commuting

8% of respondents work remotely, with about an equal number of happy and unhappy responses (44% average, 28% happy, 26% unhappy). Only 10% of those with a long commute are happy.

Coding Time

There were two questions asked about coding time. The first was how often code is written during spare time, with 29% coding frequently and 28% rarely or never. Perhaps the only significant takeaway here was that only 11% of those who rarely code in their free time are happy, while 28% are unhappy.

The second question related to whether the developer wanted to write more code, less code, or the same amount of code over the next few years. I have heard job seekers cite both too much and not enough coding as reasons for looking. 1% of all respondents reported both happiness and a desire to code less (or not at all) in the future. Over 1/3 of respondents want to write more code, compared to 17% who want to code less.

Joel Test

Responses to certain questions on the Joel Test were clearly more revealing than others.

There were 14 responses that scored positive for all of the Joel Test questions, and only one of the 14 reported being unhappy. It’s clearly a small sample size, but these respondents reported to mostly be paid at market (50%), challenged (85%), code in their free time frequently or on occasion (71%), have competent managers (57%), work on above average teams (85%), and use newer technologies (64%).

As for the individual Joel Test elements:

Quiet — Only 1% claimed to be both happy and working in a loud environment, while half of those subjected to loud noise are unhappy. The differences between the happy, average, and unhappy in quiet offices were insignificant.

Tools — Companies that use the best tools regardless of price see over 33% happiness rates, and developers that use lesser tools experience 38% dissatisfaction.

Testers — 64% of your employers have testers, but that doesn’t impact happiness (for developers anyway).

Timely Bug Fixes — Just over half of you reported that bugs are fixed in a timely fashion and the developers try to make fixes before moving on to new code, but there the satisfied only slightly outnumber the dissatisfied. Teams that don’t squash their bugs see 44% unhappiness rates compared to a slim 10% reporting as happy.

Source Control — Three-quarters of you work in places where source control is taken seriously. Our data shows that good source control won’t guarantee developer happiness, but only 9% are happy in companies without source control (as compared to almost half being unhappy).

Continuous Integration — NOTE: Joel’s original questions were “Can you make a build in one step?” and “Do you make daily builds?”. I asked a true/false “My company practices continuous integration.” CI is practiced by over half of our respondents. As you might expect, our results here are similar to the source control question. Development shops that have CI in place have an almost identical number of happy and unhappy developers, but the happy are dwarfed by the unhappy at a 4:1 ratio where CI isn’t practiced.

Schedules — Half of you feel your work is kept on an up-to-date schedule, but that alone doesn’t impact satisfaction. 40% of those with poor schedules report being unhappy.

Requirements — 64% of you are not getting clear requirements. Good reqs result in keeping one-third of developers happy and one-seventh unhappy, while those with poor reqs are almost one-half unhappy and only one-tenth happy.

Interviewees Code — I was quite surprised to learn that only about one-third of you work for companies that require applicants to write code as part of the interview process. Once again, happiness and unhappiness numbers were nearly identical at employers that require interview coding. Where interviewees don’t code, the unhappy contingent is almost triple the size of the happy group.

Conclusions

I would have expected more than 18% of all respondents would have reported as happy when compared to the 30% unhappy.

Across all questions with a third “average” answer available, many or most respondents (43-74%) chose that response.

The ratio of happy to unhappy tends to be similar (and near 1:1) in several of our questions when the answer is positive (using the best tools, CI, etc.), but the ratio jumps significantly when we review the negative responses. It seems that “having” some positive characteristic in your environment doesn’t make workers happy, but “not having” it makes workers unhappy. Developers seem to have some basic expectations that, when met, don’t impact happiness. When unmet, morale becomes an issue.

This survey and my analysis has obvious flaws. Our sample is almost entirely DZone readers, an audience that may not be representative of the global development community. It was intended for fun, some insight, and to take the temperature of the audience.

One major factor in overall happiness which was not addressed is the mission of the employer. Many employees achieve job satisfaction by working for companies that share their values and principles (non-profits or civic groups). There are clearly other questions we could have asked to get a true sense of overall happiness.

4 comments

  1. Mike S.

    Did I read that right? Places without Continuous Integration tend to have happier developers? Fascinating.

    Our problem with Continuous Integration is that our software stack is complicated enough that we spend a lot of time babysitting the CI server. Commits of all kinds routinely break the CI process. So that’s a burden. But it doesn’t make me unhappy, it’s just a minor annoyance and something I take as a sign that our application configuration needs to be redesigned.

    • fecak

      Actually you read it right, but I wrote it wrong. Just corrected it, but it should have read that the happy are dwarfed by the unhappy. Good catch!

      • fecak

        Almost 15K people read this piece over on DZone and nobody caught the error (including my editor) – so good catch! For clarity, of those that work where CI is NOT employed, 11% are happy and 39% unhappy. Where CI IS employed, the happy and unhappy are virtually equal.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s