Productivity of software development teams is difficult to get a grip on, but virtually everyone wants teams to be more productive. We often talk to managers, who say they’d like their teams to be more productive, or in their words ‘go faster’. Today we write a bit more about what we see around productivity in software development, and how seeing productivity as a property of a system can help you spot symptoms, fire-fighting and work towards more durable solutions.
- Productivity (revisited)
- Productivity is (mostly) a property of the system
- The ‘system’ is management’s job
- Sometimes it is the individual
- Productivity is an enabler for everything else
- Further reading
Measuring productivity in software development is a hard topic. What often happens in agile organizations is focusing on the effort spent, like story points. Gojko Adzic writes about this in The Play, the points, and the biggest lie in software Measuring productivity is hard, measuring value delivered is even harder.
Productivity is (mostly) a property of the system
Productivity is mostly determined by the system of work, the way teams, their dependencies and their interactions are organized. Deming stated it already many years ago that 90-95% of performance is governed by the system.
We assume people are doing their best given the knowledge, skills and tools they have at hand. Seeing the team as a (complex) system, it is safe to assume it is performing at its optimal level of productivity. If you assume otherwise, you are assuming the team members are just slacking off.
In The Gardeners Dilemma, Dirk Jan Swagerman writes about the maintenance burden you build up when you keep on delivering software. If you don’t tend to it or if you assume maintenance is free, it will drag down a team’s performance more and more, while it’s not the team’s fault - “No: your software team does not suck if they make slow progress on a legacy code base.”
Sometimes freelance developers get hired to help out with a legacy migration project, and then get fired because they are too slow, as they too struggle with the legacy and delayed feedback loops in the organization. This can happen to the best of us.
The ‘system’ is management’s job
So if a team’s productivity is not good enough, look at the system to learn what is limiting their performance. This goes beyond reacting on the first symptoms you notice, that will probably make this worse. It requires some deep, collaborative reflection using intuition support techniques like Diagrams of Effects, Cynefin, constraint mapping, value stream mapping.
A team is a complex system. This means that improving productivity is not a matter of meticulously analysing the current situation, designing a desired state and planning actions to bridge that state. Instead, the suitable approach is one of sense-making, gaining situational awareness in the direction the system is inclined to move (dispositioned), and then thinking up some experiments that are in line with what you have learned. Do experiments because you cannot be sure if they will succeed so you’d rather keep a possible negative effect small. Still run the experiment, because you never know what more you’ll discover.
This is management’s job! This is where you can have true leverage in the organization’s performance.
When I say management’s job, this is not restricted to those who have ‘manager’ in their job title, although it should at least come from there. If we succeed at having teams develop more reflective capability and systemic thinking skills of their own, they can can learn to do this on their own. So we mean ‘management’ here eventually in the sense like Gerald Weinberg uses it in his Quality Software Management series.
So you could find some things that is keeping productivity low:
- Missing skills - so they need to learn and train; this requires support and investment (training courses and time to practice; if you expect people to learn everything in their evening hours, you’re not taking productivity seriously, and you’re harming people with families, you’re depriving people of their rest, an important pillar of high performance)
- Handoffs: the team needs to wait for others outside the team, e.g. for approvals, reviews, or they have to wait on long builds, nightly test run results, etc. Waiting is destructive for productivity. It is quite useless to urge people to fill the waiting time with other work, because waiting breaks focus, people lose time in task switching. This can be a tougher one to break; create collaborative awareness of where there is waiting in the process and how that is impacting end to end (“concept to cash”) development performance, e.g. with value stream mapping. Reorganize to eliminate waiting; relentlessly improve builds and CI/CD pipelines so that they will give their feedback within 15 minutes - see Charity Majors’ post on Why are my tests so slow?
- Complacency: the team is working in a kind of ‘good enough’ mode, there is no sense of urgency. So create a sense of urgency, make sure they understand the why. In the light of a short term delivery date, a complacent team can suddenly become a focused, productive team.
- Penny Wise, Pound Foolish. Does your investment in e.g. CI/CD pipelines, developer workstations, software, webcams, headsets etc. match your wage bill, or do you have an IT service desk that says ‘No’ on every request, causing lost productivity costing hundreds of dollars per hour to save a few dollars per year on basic equipment?
- Productive individuals the team is working as a loose collection of individuals instead of a team, everyone does his/her own tasks, some are very busy, others finish early and don’t know what to do next. Invest in team coaching, teach them about flow.
And don’t focus on individual utilization! By optimizing everyone’s utilization, you are guaranteed to cause horrible development performance. Your goal is to deliver better software sooner, your goal is not to keep people busy. If you want to keep people busy, suggest them a hobby.
Sometimes it is the individual
One developer can drag down a team, by slacking off, or with a negative attitude or by working on his own instead of collaborating (being a rock star instead of a team player), or being a ‘brilliant jerk’.
This sometimes happens and this will negatively impact the team’s performance. It is still management’s job to handle this! Management hired, and hopefully, coaches the team members.
You could fix low performance by having a talk with someone who is dragging down the team, or even removing them. This does not necessarily lead to a highly productive team. The organizational context will quickly become the next bottleneck. Before acting on individuals, first ask yourself, how does the system generate the behaviour we are seeing?
So is it the individual, or is their behaviour a symptom of underlying problems?
Productivity is an enabler for everything else
Focus on improving productivity, by many small improvement decisions will create more and more time for valuable stuff. Small improvements might not change much from today to tomorrow, but on the longer term it can result in order of magnitude improvements, e.g. delivering software daily instead of monthly. And order of magnitude improvements change the game, for the team as well as for business.
- We wrote about how experiments move you forward. Experimentation is also a system, the more you do it, the lower the cost per experiment gets.
- We have more posts on productivity from management and team perspectives. Some more systemic, like the current post, others more focused on e.g. (counter-)productive use of automated tests.
- Under Pressure How to make space for improvements when the pressure is on.