This week we wrapped up the book by discussing the art of the meeting, beastly practices, and the conclusion.
The Art of the Meeting
- Lean Coffee has some of the feel of an Agile retrospective, but it’s about more than what the team is rallying around at the moment. Jameson has participated in one of these meetings.
- “We want to avoid the mistake of having a goal to keep people busy all the time when the goal should be to generate value for the business.” Geoff thought this struck a tone of mutual exclusion: People should be busy generating value.
- An example of being busy without providing value… Being asked to create a report that’s rarely used by one person.
- All three of us experienced an interesting version of daily standups at a previous company: All the teams in one room, 15 minutes to have everyone do the did/doing/will-do routine. We agreed with the author’s statements about spending most of your waiting time thinking about what you’ll say (i.e., not listening to others) and tuning out once you’ve done your part.
- The book suggested replacing the status-checking part of standups with better questions:
- What’s blocked?
- What’s about to become blocked?
- Is there work being done that’s not on the board?
- We also liked the concept of talking to the stories/tickets instead of the person working those items. You want to focus on the work, not the person.
Beastly Practices
- The book makes the case of flow-metrics including non-working days such as holidays and weekends. This didn’t seem to make an impact on us. Houston suggested that including those days depends on what kind of business you operate — think Amazon delivering packages in two days vs. five days for a bank to respond. When you’re dealing with a time frame as long as 30 days, be clear and transparent about whether you mean (1) I’ll get this to you in a month, or (2) I’ll get this to you after 30 business days.
- “High activity levels do not equate to high business value.”
- Gantt charts are more successful for project management, and for coordinating things of known size/complexity. They don’t seem helpful for dealing with estimates.
- Gantt charts also don’t account for wait/block times; it’s an ideal picture of sequencing. A common response is to put in contingency buffers, which inevitably get consumed (see also Parkinson’s law). The root problem is that estimates are treated like deadlines.
- “Instead of managing work with Gantt charts, consider managing work with queues.” We were initially confused about what this means. We think it’s about having a backlog (i.e., queue of work to do) and swimlanes to describe states (e.g., to do, doing, done).
- The author lists four reasons not to have people’s names associated with swimlanes, preferring task-based swimlanes instead:
- Focusing on the work, not the person
- Perception of poor person performance if a ticket isn’t moving
- People feel they can’t touch work that’s not theirs
- People aren’t incentivized to help others.
- Houston said he’s not worked in an environment where the last two reasons were true. If there’s work to do with someone else’s name on it, jump in and help.
- Having fewer systems of tracking work helps reduce the cognitive load of understanding what’s next.
Conclusion
- Jameson said it’s sometimes useful to consider roles as “resources.” Sure, we need to treat individuals as human beings instead of cogs. However, roles are a unit of measure (e.g., two developers, one cloud engineer).
- “Small change meets with less defiance.” Geoff is experiencing some of this with his teams that are already dealing with what they called “change fatigue.”
- “There is always more demand than capacity.”
- “Don’t punish the people; change the system.”