Boosting the effectiveness of a big development team
Hi,
As an engineering manager, I lead a great cross-functional team of 10 engineers: 3 frontend and 7 backend developers. The team is based in EMEA and we closely collaborate with our Product Manager and UX Designer, who are based in California (yes, there’s a 9-hour time difference).
One of the biggest challenges we’re facing right now is the inefficiency of our backlog refinement sessions. Imagine 13 people in a (virtual) meeting discussing a topic that only involves 2-3 of them - it’s not the best use of everyone’s time. Additionally, we typically assign 1-2 engineers per workstream, which means we’re handling 5-6 different topics simultaneously.
We assign engineers to specific topics rather than involving the entire team because we often deal with external dependencies or require extensive research before execution. Additionally, we designate technical owners for each topic to ensure clear responsibility for de-risking projects, managing timelines and ensuring a successful release to users.
I’m considering introducing a Work In Progress limit to help structure the team into 2-3 focused workstreams. However, I anticipate some challenges, including dependencies such as shared releases (we’re building a standalone app with monthly releases), on-call rotations, and a shared codebase. Other concerns include ensuring effective context-sharing between squads - especially with separate ceremonies (except for sprint reviews, which likely make sense to keep joint) - and managing team members’ workload allocation.
What are your thoughts on this? How do you manage development teams of 10+ engineers?
In Scrum you don't. You encourage people to self-manage.
Think about how you might facilitate a session so people can self-organize into teams. Time-box it, with rules around minimizing those troublesome dependencies and having enough skills in each team to create a Done Increment of work, with no more than about ten people in each team.
Make it clear that teams will then be expected to self-manage their workloads and any integration dependencies they may yet have.