Delivering Projects the Pivotal Way
(doing what works)
Aaron Severs, PM @PivotalLabs
About Pivotal Labs
“Transform the way the world builds software”
Founded in 1989 - knowledge leaders in agile
Startups and enterprises co-innovate with us ← Consulting Biz
Pivotal Tracker (top project management tool) ← SaaS Product
Now 8 offices, 500+ Pivots
100% TDD + Pair Programming + XP Agile
In all offices!
San Francisco, Palo Alto, New York, LA, Toronto,
London, Denver, Boulder
Also - want to try Pivotal Tracker?
Consistently applied process
Small user stories
Pairing + TDD + CI
Who’s who on a Pivotal Project
Client PMs and Engineers are
encouraged to join the team
Writes user stories, owns the backlog & priorities,
accepts delivered stories. Defines guardrails.
Supplies UX vision, assets, and research
Responsible for technical execution; process
Deliver stories + tests
● Co-located (or pairing remotely)
● No politics / assume good intentions
Everyone shows up on time, and is on the
same (general) schedule.
When starting a new project, we always begin with an
inception - we let it take all day.
1. Building a brand new product offering / MVP
2. Major Feature / Next Phase
3. Plumbing / Infrastructure
Typical Inception Agenda
1. Overview from main stakeholder / Product Manager
2. Goals & Success Metrics
3. Risks / Risk Mitigation
4. MVP Workflows / Epics
5. User Story Scoping
6. Next Steps
Have them when you need them.
Limit full-team meetings to:
● Daily Stand-up
● Weekly IPM (iteration planning meeting)
● Weekly Retro
A Typical IPM
1. PM reviews unestimated user stories, and answers any questions
related to them
2. For each story, an engineer volunteers to describe how they might
implement the story
3. The team estimates the story with story points
● Goal is to always have enough stories estimated to take the
team through the next iteration (one week)
4. Engineers prioritize chores - PM (mostly) deals with it
Agile → Flexibility & Predictability
● Iteration: A unit of time to measure velocity
● Velocity: Total story points delivered in an
iteration. Usually use average of past 3.
● Volatility: Measures predictability
(std deviation / mean velocity) * 100
Building software is a process of continuous iteration
User Feedback / Business Needs
Happens all the time
MAY happen periodically.
The only part of the process
that may not be continuous.
Pivotal Labs flavor of Agile emphasizes continuous work and
delivery, which is different from sprint-driven / traditional “scrum”
○ Scope-based release
○ Date-based release
○ Scope & date-based release (does not exist!)
PM plans small, frequent releases by identifying them in the
backlog, but the team can (generally) release any time.
As a PM, I want a guide for writing
great user stories,
So that my team will love me.
A Good User Story
Title: Sales Rep should be able to download a Proposal as a PDF
Description: As a sales rep, I want to be able to download a PDF for a
proposal, so that I can send it to a prospect.
Given I visit the proposal summary page
When I click the “PDF Download” button
Then A PDF file is downloaded to my computer
Resources: Example PDF, showing the desired format and all fields to
be included; A mockup showing the PDF Download button
Checklist - Is this a great user story?
❏ Is the persona or user type clearly identified?
i.e. the Sales Rep
❏ Does the story have a clear beginning and end?
i.e. Sales Rep starts on the proposal summary screen and ends with a downloaded PDF
❏ Does the acceptance criteria satisfy the persona’s goals?
i.e. I want to… so that I can email it to a prospect
❏ Are resources attached to describe all (non-obvious) details
important to the business and the user?
i.e. an example PDF to show the specific format and specific set of fields to be included
❏ Does the story represent the smallest amount of verifiable
functionality that provides incremental value?
CI / TDD - your key (silent) partner
● TDD - Red, Green, Refactor
○ Creates a bias towards more tests than you would
ever imagine you need
● CI always run on push - fixing broken builds
takes precedence over everything else
● Fewer bugs, ability to release without
regression, → velocity, predictability