Yahoo! Sports: Sprint 100 & BeyondKeith Nottonson
• Unique adaptations of a long running Scrum team ‘in the wild’ and the transition to a more Lean process
• The human and technological lessons associated with ever more frequent code deployments
• How co-located events for distributed teams promote innovation and revitalization
• The importance of internal data quality in order to deliver classes of services while reducing time to market
A pitbull, a peacock, a lemur, two bears, twoelephants, a golden retriever, a dodo bird, badger, ashark, a liger, Godzilla, and a cigarette-smoking chimpwith OCD and a gambling problem walk into a bar…
Team: 10 dev, 3 qa, 2 ops, 3 pmLocation: Sunnyvale & Santa MonicaStandups : MWF 11 AMIteration: 1 week cycle, planning TH 11 AMSprint Backlog: Prioritized, no estimatesVelocity: Tickets & Time to MarketSports Dates
DAILY RELEASES DURING THE WORLD CUP
Better flowSmaller PiecesCulture Shift
Average Time to Resolution - Site
74.3
32.0 34.830.6
90.2
13.019.2
41.2
158.6
12.4
37.6
18.2
33.4
19.2
43.4
16.0
33.8
21.4
54.4
21.8
8.311.8
25.8
17.216.4
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
160.0
180.0
Day
s
Sports Team Velocity
THE SPORTS HACKTACULAR
Tool, Product, HackTool: something used internally, little ued, little launch riskProduct prototype: the functional beginnings of an idea, will need to be iterated on later, not releasable without further workHack: new feature or implementation method, small enough to launch in the next two releases, perhaps refactoring or using new technology
THE GREAT BUG SWEEP
Number of Open Sports Bugzilla Tickets by YearOpen includes New, Reopened, & Accepted
0 1 2 8 16
165
232
740
967
1170
1257
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Number of Open Tickets by Priority
0
200
400
600
800
1000
1200
1400
1600
1800
P1 P2 P3 P4 P5 No Priority
Hypothesis
Hypothesis, by adjusting priority on open bugs to reflect duration open rather then ranked order:
• it will be easier to plan and prioritize in flight work
• the priority field can be used to represent time to fix/resolve rather than a historical priority made at a point in time long in the past
• we can communicate clearly to our partners, and teach them to communicate clearly back at us
Average Lead Time by Priority
11.916.2
39.1
49.1
38.9
55.4
118.2
126.3
13.6
20.3
44.3
53.4
0.0
20.0
40.0
60.0
80.0
100.0
120.0
140.0
P1 Resolved P1 Verified P2 Resolved P2 Verified P3 Resolved P3 Verified P4 Resolved P4 Verified No Priority Resolved
No Priority Verfied
Average Resolved
Average Verified
The Great Bug Closure
Goals of the Great Bug Sweep
• Enact true P1 prioritization
• Move from Weeks and 5 year bug resolution times to Hours and 1.5 year
• Create Flow P1 and P2 classes of service and Scrum (Planned) P3, P4 and P5 swim lanes
• Reduce open tickets to ones that may actually be resolved
You have to take risks. You have to venture beyondyour comfort zone. If something does not work thefirst time, you have to try again. If one thing does notwork, you have to try something else. You have to letthem feel pain. You have to keep it fun. And youhave to keep giving energy and love, because fallingdown hurts, but learning to run is freedom.
To learn more contact [email protected]
Top Related