Do What Works...Even if it's Not Agile
Do What Works...Even if it's Not Agile
I think I’ve come to the conclusion that ‘agile’ as we know it isn’t the best starting place for everyone that wants to adopt agile.
Join the DZone community and get the full member experience.Join For Free
I think I’ve come to the conclusion that ‘agile’ as we know it isn’t the best starting place for everyone that wants to adopt agile. Some folks, sure… everyone, probably not.
For many companies something closer to a ‘team-based, iterative and incremental delivery approach, using some agile tools and techniques, wrapped within a highly-governed Lean/Kanban based program and portfolio management framework’ is actually a better place to start.
Well, many organizations really struggle forming complete cross-functional teams, building backlogs, producing working tested software on regular intervals, and breaking dependencies. In the absence of these, agile is pretty much impossible.
Scrum isn’t impossible, mind you.
Agile is impossible.
So how does a ‘team-based, iterative and incremental delivery approach, using some agile tools and techniques, wrapped within a highly-governed Lean/Kanban based program and portfolio management framework’ actually work?
Let me explain.
First, I want to form Scrum-like teams around as much of the delivery organization as I can. I’ll form teams around shared components, feature teams, services teams, etc. Ideally, I’d like to form teams around objects that would be end up being real Scrum teams in some future state.
These Scrum-like teams operate under the same rules as a normal Scrum team, they are complete and cross-functional, internally self-organizing, same roles, ceremonies, and artifacts as a normal Scrum team, but with much higher focus on stabilizing velocity and less on adaptation.
Why ‘Scrum-like’ teams?
These teams have business process and requirements dependencies all around them. They have architectural dependencies between them. They have organizational dependencies due to the current management structure and likely some degree of matrixing.
Until those dependencies are broken, it’s tough to operate as a small independent Scrum team that can inspect and adapt their way into success. Those dependencies have to be managed until they can be broken. We can’t pretend they aren’t there.
How do I manage dependencies?
This is where the ‘Lean/Kanban’ based program and portfolio governance comes in. Explicitly model the value stream from requirements identification all the way through delivery. Anyone that can’t be on a Scrum team gets modeled in this value stream.
We like to form small, dedicated, cross-functional teams (explicitly not Scrum teams) to decompose requirements, deal with cross-cutting concerns, flow work into the Scrum-like teams, and coordinate batch size along with all the upstream and downstream coordination.
Early on, we might be doing 3-6-9 even 12-18 month roadmapping, creating 3-6 month feature level release plans, and fine grained, risk adjusted release backlogs at the story level. The goal is to nail quarterly commitments and to start driving visibility for longer term planning.
Don’t really care, this is a great first step toward untangling a legacy organization that is struggling to get a foot hold adopting agile. Many companies we work with, this is agile enough, but ideally this is only the first step toward greater levels of agile maturity.
How do I increase maturity?
Goal #1 was to stabilize the system and build trust with the organization. This isn’t us against them, it’s not management against the people, it’s working within the constraints of the existing system to get better business results… and fast.
Over time, you want to continue working to reduce batch size at the enterprise level, you want to progressively reduce dependencies between teams, you want to start funding teams and business capabilities rather than projects, you want to invest to learn.
Lofty goals, huh?
That said, those are lofty goals for an organization that can’t form a team, build a backlog, or produce working tested software every few weeks. Those are lofty goals for an organization that is so mired in dependencies they can’t move, let alone self-organize.
Our belief is that we are past the early adopters. We are past the small project teams in big companies. We are past simply telling people to self-organize and inspect and adapt. We need a way to crack companies apart, to systematically refactor them into agile organizations.
Once we have the foundational structures, principles, guidelines in place, and a sufficient threshold of people is bought into the new system and understands how it operates, then we can start letting go, deprecating control structures, and really living the promise of agile.
Opinions expressed by DZone contributors are their own.