Delight Your Customers: The #noestimates Way

Delight Your Customers:

Insanity: doing the same thing over and

Over again and expecting different results.

-Albert Einstein

The #noestimates Way

Troy Tuttle, KCP Principal Consultant, Owner KanFlow.com [email protected] @troytuttle

Rate This Talk! http://spkr8.com/t/31701

Before Lean Software Existed … (In my mind anyway)

What is this #noestimates Thing?

#noestimates means different things to different people

It started as a conversation on Twitter

The #noestimates topic sparked a renewed interest in the original promise of Agile

1st Rule of #noestimates:

If estimation works for your team or your organization, Keep Calm and Carry On.

“Estimates are a function of duration and uncertainty. Since we don't like uncertainty,

we convert it simply to duration. “

-Jabe Bloom (@cyetain) #lkna14

Estimation Challenges

“Ranges that are given as +/– a constant time or constant percent are missing the problem’s exponential nature.”

Actual durations exceeded estimates by an average 2.2 times.

Little’s statistical analysis compared developer’s estimates to the actual task durations. (for example: estimate was 4 hours, it actually took 7 hours)

“Schedule Estimation and Uncertainty” IEEE Software, May/June 2006

-Todd Little

“It’s not just good enough to double the initial estimate— some teams have found it appropriate to

multiply by 3, 4, or even Pi.”

A manager’s response…

The Estimation Game

The Software Estimation Game goes something like this:

Developer: Sure! Looks like it should take about 1 week.

Manager: Joe, would you give me an estimate of how long feature X will take?

-- 2 weeks later, feature “X” is completed. --

Manager: Um, Yeah. Joe, it took twice as long as your estimate. I told management it would take 1 week. If this happens again, it will require overtime.

Developer: Ok, I understand.

Manager: Joe, would you give me an estimate of how long feature Z will take?

Developer: Sure! It should take about 3 weeks.

-- 3 weeks later, feature “Z” is completed. --

Manager: Great job Joe! Now were making progress!

The Estimation Game

Theoretical Task Completion Time

Donald Reinertsen, Managing the Design Factory

Duration Due Date

Average

Actual Task Completion Time

Duration Due Date

Average

“Expanding Work”

Parkinson’s Law

Work expands so as to fill the time available for its completion

Nature of Knowledge Work

Manufacturing = repeatable processes Product Development = one time processes

One-time processes are non-deterministic in nature

Nature of Knowledge Work

Deterministic Approach

Project Start

Knowledge

“Complete” Planned Activity

Non-deterministic Reality

Project Start

Knowledge

“Complete” ?

What Customer Really Wants Better Knowledge

Project Managers in the middle, pulling out their hair

Nature of Knowledge Work

Idea Analysis Design Build (Dev)

Test

“Design” Phases “Execute” Phases

Idea Analysis Design Build (Dev)

Test Production

“Design” Phases “Execute” Phases

What happens when we estimate?

Customer collaboration over contract negotiation

Story Points and Velocity are abstractions that insulate the team and the business from reality

“Yippie! Our Velocity was 25 story points the last sprint. Let’s go celebrate!”

To Story Point or Not to Story Point

X X X X X X X X

?

To Story Point or Not to Story Point

@duarte_vasco

-Vasco Duarte http://bit.ly/vasco_blog

The #noestimates Playbook

• Dedicated and stable teams

• Fix time and cost constraints

• Right size user stories

• Iterate and deliver frequently

• Drip Feed or Investment Funding

• Track lead time and throughput

• Speak the language of the business (don’t use terms from physics)

• Don’t use bullet points in your slides

The #noestimates Playbook

Input Q Design Develop Test Done 1 2 1

3

1 2

9

7

8

6

4

5

Lead Time

Throughput

10

14 days from this point

5 stories per week

Upstream Planning

1. Prioritize

2. Analyze

S S S S S S

S S S S

The #noestimates Playbook

If we start a new story today, we expect to complete it in:

Probabilistic Forecasting

<= 5 days (50th percentile) <= 8 days (85th percentile) <= 13 days (95th percentile)

Delight Your Customers:

Insanity: doing the same thing over and

Over again and expecting different results.

-Albert Einstein

The #noestimates Way

Troy Tuttle, KCP Principal Consultant, Owner KanFlow.com [email protected] @troytuttle

Rate This Talk! http://spkr8.com/t/31701