Re: SPI CPI Confusion by St
St
Tue Jul 25 04:30:29 CDT 2006
I'm not sure whether you want to re-baseline to "fix" this. In some
environments baselines are nearly sacred and you shouldn't re-save them
without at least two important signatures...
Though SPI and CPI are quite handy when it comes to judging status,
they can be a little confusing when drilled down to single activity
level.
In case you have a "way earlier than planned situation" your numbers
can be plausible:
1. Actual cost reported for last week on an activity i.e. ACWP >0, say
1 developer day for 500$
2. This is part of a effort estimated to take 6-developer days (3000$).
Since you have a huge plan, you might go for a simplified
0%-50%-100%-complete crediting approach for activities that are planned
to be finished in one or two reporting periods. This gives you a 50%
for work started i.e. BCWP=1.500$. i.e. CPI=3 for THIS SINGLE activity.
Not accurate but recommend practice for huge plans with many
activities.
3. The activity wasn't planned to start on or before the status date
(rare, but it happens, that e.g. Friday is already being worked on a
task planned to start monday). I.e. BCWS=0. Now this would in fact give
you an infinitely high SPI (division by zero). MSP replaces that with a
zero. Mathematically incorrect, but easier to handle with the data
types available for programming and probably also less confusing for
most users in most situations.
So the three core earned value fields are:
ACWP=500$; BCWS= 0; BCWP=1.500$ and are in my opinion all real,
plausible and realistic. SPI and CPI indices can also be interpreted if
you know the recommended practices as background. For using CPI and SPI
as management tools in huge plans I'd recommend looking at more summary
and cumulative levels (where most probably you won't have such erratic
numbers) and not touch the baseline.
Please note: above example may or may not fit your situation - I just
wanted to describe a situation where your numbers would be very
sensible.
Sameer wrote:
> If i re-baseline the project everything works out fine...
>
> "davegb" wrote:
>
> >
> > Jim Aksel wrote:
> > > Probably what is happening is that SPI is very small. The value only shows
> > > to two decimal places.
> >
> > Yes, and that would imply something terribly wrong here. If the value
> > only shows to 2 decimals, then, by deduction, the SPI must be less than
> > ..005. If the EV is only $1000, pretty small on most projects, the
> > Budgeted Cost to this point would have to be $200,000! And, since the
> > CPI is about 3, the Actual Cost must be around $300! So he has a
> > project where he's planned to spend about $200,000 (or more) to this
> > point, has spend about $300, and has earned value of $1000. No matter
> > how you shake and bake these numbers, it tells me something is terribly
> > wrong somewhere.
> >
> > >
> > > What this is saying is "I am way behind schedule, but I am certinly
> > > accplishing a lot for the amount I have spent" If you have very little
> > > accomplishment and very little spent, the ratios might become skewed.
> > >
> > > Things that will drive these numbers: SPI=BCWP/BCWS. If you've made little
> > > progress against a big baseline amount of scheduled work ....
> > > CPI=BCWP/ACWP. If I divide BCWP by a small value for actual costs then CPI
> > > becomes artificially large.
> > >
> > > The driver to ACWP is Actual Cost of work .... that is driven by your rates
> > > on the resource sheet. However, small rates also kill any meaning for BCWP
> > > ...
> > >
> > > Also, make sure you are entering actual start/finsih dates for activitieis
> > > in the "Actual Start" and "Actual Finish" date columns. This way, you will
> > > take your EV when it is earned, not when it is scheduled.
> > >
> > > Perhaps that helps.
> > >
> > > "Sameer" wrote:
> > >
> > > > Dear All;
> > > >
> > > > I have a huge project plan and in that plan there are two tasks where the
> > > > CPI is around 3.0 and the SPI is 0.......
> > > > Is this sort of a situation possible? If yes then can anyone give me a
> > > > scenario where i can try to reproduce this?
> > > >
> > > > Regards
> > > > Sameer
> >
> >