Figure 32 Crystal ball triangular distribution

Cell I rlangjtlar Distribution

B

1 At sumption Name:

Gather the Vslcc of the Customer

I Prit I

I Püiftí I

>

jB w

à

1

A

<*

i

7.»

139Û

2ÖQ0

► ¿MW

i iojök

Min ¡M

Llkellrsl IffipJ

Max iùJcF

ÛK 1

Cancel

Enier 1 Gallery |

Cairélale... ,

Help

7 Days

Likeliest Time

12 Days

Maximum Time

20 Days

3. Each triangular distribution is "loaded" with your best estimates of three types of durations that can be projected for each task. These estimates must come from a team of marketing veterans who have experience in how long tasks take to complete using the appropriate tools. "Garbage in, garbage out" can take hold here if you aren't careful; we prefer to work with D Y = f(D X). Know your tools and how long they take to enable the completion of tasks. Triangular distributions provide samples of data to calculate task completion time three different ways:

The shortest likely time to complete a task (when things go really wellyeah, right!). The most likely time (the mean or median) to complete a task.

The longest likely time to complete a task (when things go really wrongMurphy's Law awaits!).

4. The Monte Carlo Simulation randomly samples or selects a task-completion duration from within the triangular distribution, representing each task's possible range of durations. The software running the simulation uses a random-number generator to "grab" an unbiased duration from within each triangular distribution.

5. The sum of each run through the critical path of randomly selected task durations is calculated and entered into a histogram (frequency distribution), as shown in Figure 3.3 .