Lessons learned from building software at scale

So you have read The Lean Startup. Suffered through following several agile blogs (such as this one). You think you are ready to join the cool kids and have product scrums and stand-up meetings and all that other stuff. Now you need an implementation plan.

Maybe it is time to read this post by Paul Adams on the Intercon blog. He describes some of the lessons he and his development team have learned from building software and scaling it up as the company grows. I asked a few of my contacts at startup software firms what they thought of the post and there was mostly general agreement with his methodology.

Here are some of Adams’ main points to ponder.

Everyone has a different process, and the process itself changes as the company matures and grows. But his description is for their current team size of four product managers, four software designers, and 25 engineers. Like he says: “it’s not how we worked when we had a much smaller team, and it may not work when we have doubled our team.”

Create a culture where you can make small and incremental steps with lots of checkpoints, goals, and evaluations. “We always optimize for shipping the fastest, smallest, simplest thing that will get us closer to our objective and help us learn what works.” They have a weekly Friday afternoon beer-fueled demo to show how far they have gotten in their development for the week. Anyone can attend and provide comments.

Facetime is important. While a lot of folks can work remotely, they find productivity and collaboration increases when everyone is in the same room in a “pod.” Having run many remote teams, certainly local pods can be better but if you have the right managers, you can pull off remote teams too. It appears IBM is moving in this “local is better” mode lately.

Have small teams and make them strictly accountable. Adams has a series of accountability rules for when something goes wrong. Create these rules and teams and stick by them. “We never take a step without knowing the success measurement,” said one friend of mine, who agrees with much of what Adams says in his post. My friend also mentions when using small teams, “not all resources have a one-to-one relationship in terms of productivity; we find that we that the resources we use for prototyping new features can generally float between teams.”

Have a roadmap but keep things flexible and keep it transparent. “Everything in our roadmap is broken down by team objective, which is broken down into multiple projects, which in turn are broken down into individual releases,” said Adams. They use the Trello collaboration tool for this purpose, something that can either be a terrific asset or a major liability, depending on the buy-in from the rest of the team and how faithful they are to keeping it updated.

However, caution is advised: “The comprehensive approach that Adams describes would be entirely too much overhead for most startups,” says my friend. This might mean that you evaluate what it will take to produce the kind of detail that you really need. And this brings up one final point:

Don’t have too many tools, though. “Using software to build software is often slower than using whiteboards and Post-it notes. We use the minimum number of software tools to get the job done. When managing a product includes all of Google Docs, Trello, Github, Basecamp, Asana, Slack, Dropbox, and Confluence, then something is very wrong.”

Leave a Reply

Your email address will not be published. Required fields are marked *