SCRUM, two months (or years?) into it.
I’ve been part of two SCRUM rollouts so far. One was a grassroots effort, the other organizational. I have some thoughts I’d like to share, in no particular order.
- It’s no silver bullet. Repeat that last sentence 10, 50, 1000 times until you believe it. If your work habits, talent, training, and support roles suck, your (insert work product here) will still suck using SCRUM. On the bright side, you’ll understand that in 2 weeks to 2 months, instead of years after the product is in the field
- SCRUM teams: 10 people max, and that’s pushing it. Five is better. I was on a 6 person SCRUM team, and it was great.
- Team of Specialists versus Team of Generalists? I’ve no idea. I’ve arguments for either approach, and no one answer fits everywhere. Yay, “It depends.” I could be a consultant.
- The team must see the value in SCRUM. It’s a lightweight process, but it requires them to think and communicate on a level they haven’t before. If it’s a drag AND they have to communicate to one another, they’ll abandon both.
- Don’t B.S. the acceptance criteria. If you don’t complete a user story, you don’t get credit, period. This sort of boundary helps focus the SCRUM team when it’s committing to a story.
- Enable the team to organize itself once it’s gelled. Yes, that means kicking the dead weight overboard, too.
- Empower the SCRUM master to do his/her job. Measure if s/he does it.
- Project Manager != SCRUM Master. The skillset and attitude is different.
- For heaven’s sake, DO NOT MAKE THE TEAM LEAD THE SCRUM MASTER. I was a team lead and a scrum master for about 2 weeks. Disaster ensued, and I got help for my team. Basically, anything that lets the team lapse into a parent/child relationship will destroy a SCRUM team–they must feel like they have the power to make something happen.
Some things that I don’t think SCRUM wholly solves (nor seeks to, if you seek a cop-out):