Presented by Adam Yuret & Troy Magennis
"When will the project be done?!" This single question has created more dysfunction and psychological danger than probably any other single demand in history of work.
Story points, ideal hours, task hours, I’ve even seen a director who un-ironically tracks "developer minutes" to handle these issues.
Focused Objective has created some simple, easy to use (and free) forecasting tools which we can learn about while actually doing probabilistic forecasting using pen and paper.
We workshopped three scenarios:
- Reliably capturing and spotting errors in historical data
- Estimating total project size (story count) by sampling a subset of all features or epics
- Forecasting completion date using probabilistic forecasting (Monte Carlo) of estimated or measured teams’ throughput (completion rate) or velocity (points)
In this workshop you
- Learned how much sample data is required to undertake a reliable forecast
- Learned how to spot erroneous data or data that will mislead a forecast
- Learned how to use story count estimates on a subset of features to forecast a projects combined total story count, or to see if the count you have been given is likely
- Learned how to use historical data to perform a feature completion date forecast, or to see if the date you have been given is possible.
The processes described involve using dice to simulate uncertainty in projects and building a probabilistic picture of the more likely outcomes (often called Monte Carlo simulation). It is a fast and accurate way to combine historical data into meaningful and verifiable results. By performing a Monte Carlo forecast by hand, you will realize how easy the technique is to perform and not be afraid to use it in your next estimation or forecasting task.
We will also discuss the choice between estimation in points versus throughput and how this impacts forecasting accuracy. We will also discuss how most tools available for forecasting go wrong, and how to understand how accurate your forecast using these methods should be considered.
- Learn how much sample data is required to undertake a reliable forecast
- Learn how to spot erroneous data or data that will mislead a forecast
- Learn how to use story count estimates on a subset of features to forecast a projects combined total story count, or to see if the count you have been given is likely
- Learn how to use historical data to perform a feature completion date forecast, or to see if the date you have been given is possible.
Let’s bring real empirical data to the table and use it to make better decisions about how we manage our systems.
Adam Yuret is an experienced systems thinker who has consulted small non-profits and fortune 100 clients on adopting context-driven systems to solving difficult problems. Adam started Context Driven Agility in 2010 to share his passion for humanistic flow-based systems full time. He’s been consulting organizations and teams to adapt to their respective contexts using collaborative approaches and lean principles to great effect. Context Driven Agility is about recognizing that no one set of standardized best practices is applicable to every situation. Effective learning organizations strive to understand context before applying approaches or suggesting experiments.
Troy Magennis is an experienced IT executive who has been involved in many leading software organizations over 20 years. Most recently, Troy founded Focused Objective to build and promote risk management tools that simulate and forecast software development projects and portfolios. Technology has always been a passion for Troy. After cutting his teeth on early 8-bit personal computers, Troy moved into electronic engineering, which later led to positions in software application development, architecture and management for some of the most prominent corporations in automotive, banking, travel and online commerce. Email: email@example.com
The tools discussed in the meeting are available on the Focused Objective web site: