So I
lecture quite a lot about different aspects of Project Management from requirements
gathering, pre-studies, business case development to certification preparations.
And I have found myself telling the students about all sorts of things a PM has
up the sleeves . These are not things that would be found in ant text book, in
any higher moral ground theory or academic paper. The number crunchers and
process addicts would probably like to take my head off but, it’s now on the
internet so it will not go away.

These are
tools based on experience and spending a lot of time getting dirty hands. The
idea is that they will extend the life expectancy of the PM’s mental health and
probably take some of the stress off the mind of the PM. So am I giving away
the illusionist’s tricks? Am I showing the mirrors and pointing out the strings
that make the lady appear to fly? Sure, I am. But it’s not like no one else
know them… Every experienced PM knows them, they are just not telling the new
guy on the subject.

Since I already
tell people in class, why not let the NOC-list into the open? Don’t know what the
NOC-list is? Have a look at the first Mission Impossible film.

I’ll start
with PERT-estimates, or Three-point-estimates. Let’s start with the basics.

When we
need Estimates, we reach out to the subject matter experts (SME’s) and ask them for
estimates on different task. We do know that most people hesitate to actually
give a number. For self-preservation if nothing else. They have been around
before and they know that this number pretty easy turn into gospel and they don’t
like that. The three-point-estimate is where we let them say what they think
but also tell us the best and worst case scenario. So we cut them a bit of
slack and say that we will crunch out the numbers afterwards.

I’m not
going to go through the whole theory here since that is common stuff and if
anyone need to check it out it’s on
Wikipedia http://en.wikipedia.org/wiki/Three-point_estimation.

So
according to this we calculate Max + Min + 4 x Expected and divide that with 6.
So where does 4 and 6 come from? Well it is based on mathematical theory for
bell curves and standard deviation. What happens if we, shame on us if we do,
would actually change these numbers? Let’s say 3 and 5 so we have Max + Min + 3
x Expected divided with 5? What happens then and why would we do that?

So we start
with what happens. Let’s look at some figures for example. We have an imaginary
project where we came up with Min = 70 days, Max = 242 days and Expected = 125
days.

Using the
official numbers we get a PERT-calculated value of 136 days (70 + 242 + (4 *
125))/6.

We also get
a Standard deviation of 29 days. (242 – 70)/6=29.

And we get
a error percentage of 4%. (Standard Deviation / (Max + Min + 4 x Expected))

Let’s now
change to 3 and 5. What numbers do we get then?

Calculated
value = 144 = (70 + 242 + (4 * 125))/5

Standard deviation = 34

Error percentage = 5%

Standard deviation = 34

Error percentage = 5%

So what
does this mean? Basically it means that one let the deltas for the extremes
reflect more in the end calculated result.
We can see that the calculated value increased from 136 to 144.

The million
dollar question, when would you do this?

Ever get
the request from a customer “We like a fixed price project”? When doing a fixed
price project we have to carry the risk. So we like to know if this is a high
risk project or not. If the 4/6 calculation is not very far off from the 3/5,
well we are pretty sure in our estimates. If they are far off, we probably need
to break things down further and analyze more. We probably are going to use the
3/5-calculation instead when writing the proposal and we probably are going to
find the tasks with large deltas and use them as inputs for a risk resolution
budget post.

Can we increase the numbers? Say 5/7. Yes, we can and that would mean giving more leverage to the expected. So we might work with these numbers depending on the level of how certain people are in their estimates. Large deltas, probably want to reflect them more.

Can we increase the numbers? Say 5/7. Yes, we can and that would mean giving more leverage to the expected. So we might work with these numbers depending on the level of how certain people are in their estimates. Large deltas, probably want to reflect them more.

## Inga kommentarer:

## Skicka en kommentar