scrum_blr 10th meet up 13 sept-2014 - the slippery slope from agile to scrum fall - avinash rao
TRANSCRIPT
©2013, Cognizant
Paved with Good Intentions:
The slippery slope from Agile
to Scrum-Fall
Avinash Rao
The Session Today
2
Belief in the Agile Tooth Fairy is eroding
The Fall of Waterfall, and Agile Success
Scrum-Fall
Lessons Learnt – Beyond the Case
The Empire Strikes Back
Act 1
Act 2
Act 3
Epilogue
Prologue
Scope based funding
“Do Agile” for the same approved scope and budget
Use outsourcing partners to cut costs – have Scrum teams on tap
“Hmm, so you need a Process Owner from the Business? We are very Busy, but lets see … ”
“Go Agile” makes it easy for programs to J<expletive deleted>DI
We’ve gone from not accounting for evolution and emergence, to making
it up as we go along, with the funding process still asking for predictable Progress
… but “Management” hasn’t changed
6
A committed scope (backlog) of functional items, delivered with dispersed teams and SMEs representing the Business; overlaid by the usual corporate IT governance processes
Leading to:
'We'll figure it out as we go along' Architecture
Evolution in corporate security and IT governance guidelines
The Agency problem between business and the representatives of the business
The temptation to build functionality quickly to demonstrate progress
What We Get when we Go Agile
7
High visibility Board-reviewed NPD program
~100,000 FP Product Suite
50% budget spent, 25% scope completed
Design Silos – caught only at integration test
At risk of scrappage for the second time!
The Situation
10
Requirements phase that takes forever
and then some …
Supporting applications built, and approaching obsolescence
Search for the Perfect solutions
The Usual Suspects were in Play
11
Finite scope for 18 month period
Release to business POs, no production releases
Common governance for 3 IT vendors
Continuous Integration
2 week iterations
One Team, common tooling
Full time DevOps team
Clear DoD, with defined Coding and Test coverage standards
The Move to Agile
12
3 fold Productivity gain / FP delivered
55% cost reductions / FP delivered
60% reduction in defects slipped out of Dev
On-time deliveries every 2 weeks, integration tested
Agile Awards in UK and USA
The Results
13
But Business wasn’t entirely pleased
As all IT Managers know, this was because Business is Evil
So, IT achieved a famous victory …
17
Original scope was not a full replacement for current system in Production
POs had validated Use Cases, not end to end Business Process flows
Fit and Finish issues – anything not in the DoD was inconsistent
However, additional funding provided to complete the product; the Business teams took direct control of the project – IT SMEs replaced by
Business POs for direction setting
Alternate Theories for Business Displeasure
18
POs now set scope for each release
19
1 2 3 4 5 6 7 8 9 10 11 12
Rework increased 72% when backlog set by POs for each iteration
Designated Process Owners rarely have financial control (or
sometimes even an understanding of the overall business case
and trade-offs) needed to balance progress on improvements
v/s the funding to completed committed scope and Quality.
IT vendors instructed to increase productivity
In return, iteration timelines were
doubled
The BA doesn’t contribute to code velocity, why not replace him with a developer?
Higher Rework Increased Cost Pressures
20
80
90
100
1 2 3 4 5 6
PQ
Decreasing PQ Targets
Defects Up, Productivity Down
21
0 5 10 15 20 25 30
0 5 10 15
As defects ballooned, teams were
overwhelmed by defect fix effort
Productivity became erratic, and
decreasing (12 month view)
Attrition increased, leading to ….
Scrum-Fall
Fixed Scope
Committed, fixed scope to be delivered every iteration
Game-playing with Buffers to ensure scope committed is
achievable
Additional Teams and Stages
Additional Test Teams, Validation teams …
Added layers mimicking
traditional QA
Measures to increase velocity actually decreased
velocity
Defect fix effort a major drag on productivity
Decreased Velocity
23
Funding Agile Projects
25
Phase 1 complete
CRs on Phase 1
Acceptance
Project completion Benefit realization
Project size: 100 FPs; Work done equivalent to 180 FPs
Measured by size (often the basis for funding), Agile projects show a
slower rate of progress because of rework – this rework would have
been funded by CRs on traditional projects
Test the Business Process, not the Use Case
26
Business Process
Pass/
Fail
1 Achieves A
2 Achieves B
3 Achieves C
4 Achieves D
5 Achieves E
6 Achieves F
UC 1
UC 2
UC 3
UC 4
Though UCs may Pass testing, Business Processes must Pass for Business
acceptance
Fixing later costs as much as Right-First-Time
27
Agile provides a seemingly easy chance to change and improve later … this is not true for NFR and Architectural items
100%
150%
300%
0%
50%
100%
150%
200%
250%
300%
350%
Original cost of
component
Cost of 80%
improvement
Cost of an
additional 10%improvement on
original NFR