A Different Kind of Delivery

In my day job I’m a software development consultant, which means I work in a delivery team and spend a great deal of time working closely with my product owner to build great software. Outside of my day job, I was recently part of a big project completely unrelated to software and was on the other side of the fence, as the (co-) product owner working with a rather different type of delivery team.

It has been a huge learning experience for me, not just getting the product owner’s perspective but also observing how delivery teams work in other industries. In this post I’ll share some of the most important lessons which I think we can carry across to software development.

I’ll start with a quick overview of the project. It started with a roughly forty-week development phase, then a very intense production deployment, and continues now with an ongoing development-in-production phase.

The “product” is my son, delivered in September last year.

The biggest lesson has been a pretty significant rethink of project scope, must-haves and nice-to-haves. It turns out the list of must-haves is much smaller than the feature list of a typical adult human, basically just breathing, feeding, sleeping, and waste disposal, with a fairly primitive alerting system which doesn’t even really distinguish between which of the above four functions has an issue.

The ability to walk and talk are two of the most critical skills for humans—what I would generally consider as must-haves—but my son wasn’t born with either of these features and even now at seven months of age they’re still a while off. There are good reasons that these features are deferred until after the production deployment.

Firstly, they take a long time to develop, and technically you can survive without them, so insisting they be ready for the initial deployment really only serves to delay the initial deployment (not a pleasant thought for my wife). But secondly and more importantly, they would be too hard or likely impossible to develop well without a feedback loop, which is the next big lesson.

Real feedback loops are critically important to developing most features. The processes required to walk and talk are just far too complicated to be developed in isolation without the assistance of good feedback mechanisms. They take hours and hours of repetitive trial-and-error to fine-tune, and no amount of deep thought, design or planning will be an adequate substitute for this.

Note also that I said real feedback loops. Many development teams will use lower environments to test and iterate on features, and this is certainly better than no feedback loop at all, but sometimes there really is just nothing like production and no tester like a real user. You can still do this in a controlled way without releasing untested features to the entire world, through techniques like feature flags and deployment rings.

My son can practice his walking and talking safely at home before he progresses to the outside world, but even then, as the saying goes, “you have to crawl before you can walk”, which brings us to the third big lesson.

Iterative development can mean delivering temporary, interim features that won’t be needed in the long-term. When time is short and you’re already being ruthless with your must-haves and nice-to-haves, developing features that will be removed or replaced later can seem like a waste of time. But some features are just too big to implement in one piece—whether that be within a release or a sprint—and it makes sense to instead implement a smaller or simpler feature that will start the feedback loop sooner.

You can then iterate from this interim solution by changing or extending where possible, or at least by applying lessons learnt from the feedback loop to deliver with more confidence. There’s another saying for this, “make it work, make it right, make it fast”.

The fourth lesson was about staying flexible and adapting to change. So much of what happens, before and after the birth, is out of your hands. It’s good to have a plan, but it’s better to have a plan that’s flexible. You are encouraged to decide on a birth plan, but anything can happen on the day so you need to be ready to adapt. Similarly with all of the various clothes and supplies you need to by—we found it was best to buy conservatively at first, see what works and what we like, and then buy more as needed. It’s hard to know what you need to pack in your nappy bag until you need it, so the contents change over time as you encounter more scenarios.

One lever you can use to help adjust to change is the length of your feedback cycle, like your sprint length. Early on in the pregnancy when things develop slowly, we would only see our obstetrician once per month. This eventually increased to once per fortnight, once per week and then several times per week right before the delivery, as things began to change faster and faster. After the birth you have regular visits with a maternal and child health nurse. The frequency of these visits follows the opposite pattern, from once a fortnight, once a month, to once every several months, as development starts to stabilise and slow down.

Some things just can’t be planned at all. Even ignoring the ethics of it, it would be foolish of me to try to plan out my child’s hobbies and interests. For some things, it’s best to just provide a lot of opportunities for experimentation and then invest further in the things that work.

The fifth lesson was on conducting effective handovers. In consulting its quite common for the delivery team to hand the application over to another team at the end of the project, like an in-house development or operations team, but this can also happen at product companies where a development team hands an application off to an operations team. This usually involves some amount of documentation and a knowledge transfer workshop.

I’ve never found this format to be particularly effective, as it’s extremely difficult for somebody with no context of an application to know what questions to ask or which points to pay particular attention to.

After the delivery of my son, my wife and I stayed at the hospital for four nights and were the primary carers, but we had 24/7 access to a team of midwives and nurses who walked us through everything we needed to learn and could be called upon at any time to assist. Four days is long enough that you’ll need to run through the main processes (feeding, nappy changes, putting to sleep, bathing) several times, and will need to diagnose a few outbreaks of crying. This means that the handover doesn’t end until you have demonstrated the ability to manage on your own.

I always think it’s worthwhile looking to other industries to see what lessons can be applied to software development. Many of the problems we face are more universal than you’d think, and it’s interesting to see the approaches others have found to dealing with them. As you deal with other professions in your day-to-day life, keep an eye out for things you can learn from them.