How we used Kanban in Operations to Get Things Done
summary
Ops team runs 27 million players/day, must stand up six data centers in six countries in six months while handling live incidents; ticket backlog shows arrivals outpace closures, a long tail with 21% of work taking >120 days, and mode at 1–2 days. Introducing a physical Kanban board, WIP limits (experimenting from 40 down to WIP limits to 10 then 5–7), explicit policies and work item types, and a CAB queue cuts validate to three days, yields 10 finished pilot projects, and reduces context switching and cycle time.
AI-written summary · about
chapters (23)
- 0:00Introduction
- 1:20Organizational restructure
- 2:30What is the demand
- 2:50What is the lead time
- 3:13What is the cycle time
- 3:29How we got started
- 4:40Why do people do that
- 6:00What is hourly time
- 8:17Where work gets stuck
- 9:11Meeting upstream customers
- 10:11Working in progress limits
- 11:03Reorg
- 12:08Board
- 14:28Queues
- 20:02Work Item Types
- 20:58Whip Limits
- 22:21Email Example
- 22:41Continuous Improvement
- 24:29Death March
- 26:02The Leaders Task
- 26:28The Workers Task
- 26:44Book List
- 27:21Questions
