What we'll cover:

• What is task/project estimation
• The three-point estimation
• Other considerations

A good project is a project that delivers:

• Expected features
• On time
• On budget
• With expected quality

• As accurate as possible
• Not too optimistic
• Not too pessimistic

We all have cognitive bias

• Wishful thinking
• Anchoring / Focalism
• Planning fallacy
• Cognitive dissonance

• How complex and big is the project ?
• Have we ever made something like this ?

... cut it in consistent tasks.

Take three figures (points):

• b = the best-case estimate
• m = the most likely estimate
• w = the worst-case estimate

Estimation is a weighted average.

e = (b + 4m + w) / 6

And standard deviation

σ = (w − b) / 6

• b = 2d
• m = 4d
• w = 10d

e = (2 + 4*4 + 10) / 6 = 4,666...

σ = (10 − 2) / 6 = 1,333...

Use SD to add confidence level:

• Confidence level in E ± σ is 68%
• Confidence level in E ± 1,645σ is 90%
• Confidence level in E ± 2σ is 95%
• Confidence level in E ± 3σ is 99.7%

You can estimate in group:

• It reduces bias
• It adds more expertise
• It involves the team (which is good)

To estimate a project based on tasks estimation (e):

E = Σe

SD = √(Σσ^2)

The more subtasks you estimate, the more precise your project estimation will be.

• Ressources are never 100% productive
• Context switching has a cost
• Management time, meetings and contingence has a cost too
• Cognitive bias tend to make estimations too optimistic
• Those who estimate are not necessarily those who are going to do

Effort and costs are different from delay / planning

(but you have PERT to solve that)

• Monte Carlo Simulation
• *Function points
• COCOMO / COSYSMO
• Story points and Planning poker
• PROBE
• Use Case Points
• Putnam

and a thousand more...