Agile Project Management at The Washington Post

Innovation through Iteration

Background

Background

Washington Post IT Unit

• About 140 people

• Supports operations of the newspaper and some operations at other

Washington Post Company affiliates, including:

• Publishing

• Advertising

• Circulation

• Syndication

• Accounting

• Production

Washington Post Web Solutions

Traditional projects flowed like a waterfall

htt

p:/

/flic

kr.com

/photo

s/2

4028533@

N0

3/2

297

19

07

95

/

• Test Scripts

• Working Build

• Wireframes

• Architecture Diagrams

• Requirements Doc

• Known specs

• Discrete phases

• Tight discipline

• Specific and unchanging requirements

• Design and development standards

The Waterfall: Measure twice, cut once

Discovery

Design

Development

Testing

Deployment

• Launch

The goal: Build the thing right.

• When it's familiar territory

• Involve high levels of integration with existing

systems

• When working prototypes for user feedback are

more expensive or difficult to produce (e.g.,

non-web)

Waterfall works well for large-scale projects

Waterfall projects

Familiar territory! !Integration with DSI!Simple transactions

Waterfall projects

Familiar territory" Simple transactions" Integration with PAS"

• Simplified project governance (Senior Management)

• Bigger projects mean fewer per year to track

• Fixed scope keep stakeholders/project team aligned

• Hoarding of IT resources through project bloat (at Discovery)

• More risk that changing business needs will outpace development

• Inaccurate LOE and schedule estimates

• Tendency toward "Launch and move on" mentality

Potential effects of waterfall projects

Effect on project managers

Effect on project managers

When things go wrong in the waterfall

“By the time the project finished, the business needs had totally changed.” – Business Analyst

“By the time the site launched it looked completely different from what we had envisioned.” – Designer

“If I knew in the beginning what I know now, we would have made a very different site.” – Business Client

“We had to cut some corners – documentation, user testing, support training – but we made the date.” – Project Team

“Hey, you approved it.” – PM

Knowledge gap when building unknown solutionsvo

lum

e

decisions

knowledge

Discovery

Design

Development

Testing

Deployment v1.0

So that was then. . . .

16

Business in transition

Business in transition

•“We actively develop new revenue streams from non-traditional

sources.”

•“We introduce and support new brands, selectively, when we

believe that doing so allows us to achieve the full potential an

opportunity may afford.”

•“We make bets on ideas that can have material impact even if

they entail high risk. We invest in small-scale experiments to

learn more about areas of strategic opportunity where

uncertainty is high. More than ever, responsible innovation is

necessary for our success.”

•“Because of the high level of marketplace uncertainty, we

regularly monitor and revisit our strategies, being willing and

Strategic Focus: Innovation

Where IT comes in

Align our methodologies to support innovation. . .

• Partner with the business to explore and realize new revenue streams

• Enable those “bets” and “small-scale experiments”

• Improve speed to market; bring value faster

. . . While we remain true to our core mission of supporting the

traditional business

A shift in emphasis

Waterfall:

Iterative:

Build the thing right.

Build the right thing.

An alternate approach: Iterative

Discovery

Design

Development

Testing

Deployment v1.0

T I M E

An alternate approach: Iterative

ß ß ß ß ß

T I M E

• Better fit for product innovation

• Speed to market with beta releases

• Betas prove/refine the concept

• Earlier value generation

• More user feedback, which guides the next iterations

The goal: Build the right thing.

Iterative vs. incrementalhtt

p:/

/ww

w.fl

ickr.com

/photo

s/s

pie

lzim

mer/

42

92

15

17

2/

Got the whole brick wall metaphor from Jeff Patton talking to Jared Spool.

http://www.uie.com/brainsparks/2008/08/05/spoolcast-ux-in-an-agile-environment-with-jeff-patton/

Why Vine?

• Looked at many topics

• Beer/Wine/Spirits meets three criteria:

• Consumer Passion

• Advertiser dollars in market that we don’t get

• Consumer spending

• Negatives

• Variation in state laws (MD/VA/DC)

• Local retailer needs

• Is the universe of local wine lovers large enough?

Initial idea

Pilot project: Vine

The Vine Betas

• Registration

• User Profiles

• Coupons

• Credit Card

• Wine organizer

• Chat

• Social

Networking

• Product blog

• Program info

• Vine content

• Analytics

• Monitoring

• Customer feedback

The Vine Betas

Scrum training and adoption

Scrum roles and deliverables

• Product backlog

• Single product owner on the business side who is part of the team

• 30-day sprints

• Self-managing teams

• Technically, no project manager. Instead, ScrumMaster.

• Streamlined documentation

Product backlog

Sprint task list

System Documentation

Member home page

Featured Selections/Discounts

Member Content/Events

“My bar/cellar”

Store/Retrieve bottles

Thoughts/Lessons so far

40

• Do you ever get the feeling

that you’re surrounded by

total and complete chaos?

• Organizational inertia, cultural

change

• Integration with enterprise systems

• Transition from Beta to bulletproof

• Abandoning unsuccessful Betas

Challenges/Risks with iterative products

Sprint zero is critical.

http://www.flickr.com/photos/ayelie/441101223/

http://www.flickr.com/photos/42dreams/2452861482/

http://www.flickr.com/photos/cursedthing/448971179/

Not every iteration is a public release

http://www.flickr.com/photos/basykes/34279052

http://www.flickr.com/photos/kellysue/2831068087/ http://www.flickr.com/photos/clevercupcakes/3229153310/

http://www.flickr.com/photos/bossacafez/268979524/

Not every iteration is a release

http://www.flickr.com/photos/basykes/34279052

http://www.flickr.com/photos/flirtykitty/9226535/

Modular code enables re-use

ß

Social

Networking

ß

Shopping Cart

ß

Credit Card

Processing

ß

Mobile

Browsing

ß

Text

Messaging

ß

Google Maps

Integration

ß

Video Player

ß

Rating/

Reviewing

• When the feature set is evolving

• Bets on ideas; small-scale experiments

• Minimal IT investment

• Low-cost failure

• Because it’s in line with the advantages of the web

• Easier to update, enhance, evolve

• Instant customer feedback

• Incremental releases of new functionality (Betas)

• Product improves as more people use it

Iterative works well. . .

That’s it.

47

47

Dave Burke

[email protected]

twitter.com/daveburke

Contact