Agile Experimentation in Everyday Life - A Guide to More Aha! moments by Miloš Belčević

In the realm of software development, the pursuit of accurate project estimation has been a perpetual challenge. Despite the rise of Agile methods, the specter of inaccurate estimation continues to cast a shadow over projects, leading to missed deadlines, exceeded budgets, and frustrated stakeholders.

Consider this: a groundbreaking study by the Standish Group in 1995 revealed that a staggering 31.1% of software projects were canceled before completion, with over 50% ending up costing nearly twice their initial estimates. Fast forward to the present day, and while Agile has undeniably transformed development practices, the issue of project estimation remains a thorny one.

While Agile projects boast a commendable success rate three times higher than traditional waterfall approaches, over 50% still grapple with time and cost overruns. The question then arises: why has Agile, with its iterative approach and emphasis on collaboration, not completely eradicated the problem of inaccurate estimation?

Agile introduced relative estimation, epitomized by story points, in contrast to the upfront man-day estimations of the past. However, the journey towards accurate estimation has been fraught with challenges. Despite their widespread adoption, story points have often fallen short, leading to counterintuitive outcomes. A case study within a prominent corporation revealed that stories rated lower in complexity took longer to complete than ostensibly more complex ones.

This dilemma underscores a fundamental truth: the challenge of estimation transcends estimation methods; it is deeply rooted in human nature. Our innate biases and tendencies toward optimism color our estimations, rendering them prone to error. To break free from this cycle, a paradigm shift is necessary—one that embraces a data-driven approach.

The answer: actionable agile metrics and probabilistic forecasting. By leveraging historical data, teams can move beyond guesswork toward informed decision-making. These metrics provide nuanced insights into team performance and project dynamics, empowering teams to make accurate predictions about future outcomes.



During this talk/presentation, I will share:


- the results of two studies by the Standish Group (1995, 2020)

- a case study about story points from one US corporation

- what metrics we need to gather as well as how (and why)

- some cool models and tools (through quick demos or screenshots)

In this illuminating talk, we'll demystify agile estimation, drawing from real-world examples and personal experiences. Attendees will gain practical insights into the tools and techniques that underpin effective estimation practices. By the end of the session, participants will be armed with actionable strategies and newfound knowledge to navigate the estimation challenge confidently, ensuring smoother sailing on their Agile journey.

1. Dimitrije Davidovic
Never Miss Your Deadline Again:
The Future of Forecasting Is Here!

3. Dimitrije Davidovic
Scrum Master, Agile Consultant,
Professional Kanban Trainer
ADD IMAGE
ProKanban.org

4. The Estimation Problem is as old as
Software Development itself

5. CHAOS Report (1994)

6. Why?

7. Fixed Mindset
“Plan the Work and Work the Plan”

8. “It always takes longer than you expect,
even when you take into account
Hofstadter's Law.”
Douglas Hofstadter
Hofstadter's
Law:

9. Agile to the Rescue

10. CHAOS Report (2015)

11. CHAOS Report (2015)

12. Ideal Days
Man-hours
Resource/Capacity
Planning
Gantt Charts
Estimation/Forecasting in
Traditional (Waterfall) Projects

13. Story Points
Velocity
Velocity-Based
Planning
Estimation/Forecasting in
Agile Projects

14. “Story points are (relative) units of measure for
expressing an estimate of the overall effort that will
be required to fully implement an item or any other
piece of work.”
Estimation Technique
Invented by Ron Jeffries
Part of Extreme Programming (XP)
Frequently Misused
Story Points

15. Our Customers don’t care about
Story Points…
All they (usually) care about is:
The Biggest Problem with Story Points
“When Will It Be Done?”

16. Story Points and Average Velocity cannot answer this question

17. Why?
We have 3 additional problems

18. Problem 1: Guessing Game
An estimation is just our guess
“Accurate estimate” is an oxymoron
(something is either accurate or an estimate)

19. Case Study @ Ultimate Software (2016):
Estimate
(Story Points)
Average Time
to Complete
3 points
5 points
17 days
13 days
Problem 2: Points ≠ Time

20. Problem 3: Misuse
We cannot treat numbers that represent categories as
numbers with mathematical values

21. What’s our average Velocity?

22. What we know by now:
As humans, we are terribly bad at estimating
Story Points won’t help us solve this problem
Average Velocity doesn’t make any sense

23. What can we do about this?

24. “We cannot solve our problems with the
same thinking we used when we created
them.”
Albert Einstein

25. Instead of relying on our gut
feelings and “expert opinions” that
got us here...
We need to change the game!

26. The answer is:
Using real, historical data

to get reliable forecasts

for our future delivery

27. Flow is the movement of potential
value (work items) within a process
Flow Metrics:
Cycle Time
WIP
Throughput

29. The Monte Carlo Simulation
To forecast, we try to “simulate” the past and apply it to the future
It’s a probabilistic forecast, meaning we don’t get a single result, but many options that are
associated with a probability

30. Use case #1: Release Date
Today is 17th of May
We have 100 work items to complete in our Backlog
We need to figure out our Release Date
Our Customer is asking: “What about 1st of July ?”

31. Using our historical Throughput

32. The Monte Carlo Simulation: Release Date

33. The Monte Carlo Simulation: Release Date

34. The Monte Carlo Simulation: Release Date

35. Use case #2: Number of Tasks
Today is 17th of May
Our Sprint starts Today and lasts for 2 weeks
Our PO proposed a Sprint Goal
We need to complete 40 PBIs in order to achieve it

36. Using our historical Throughput

37. The Monte Carlo Simulation: Number of Tasks

38. The Monte Carlo Simulation: Number of Tasks

39. The Difference
Traditional EstimationProbabilistic Forecasting
Based on gut feelingBased on the real,
historical data
Deterministic ThinkingProbabilistic Thinking
Human Factor (Bias)Statistics

40. Benefits of the Monte Carlo Simulation
Time Savings
More Accurate/Reliable Forecasts
Realistic Plans
On-Time Delivery
Increased Trust & Transparency

41. PLATINUM SPONSORGOLD SPONSOR
Thank you!
SILVER SPONSOR
Let’s Connect!