Best practice for modelling likelihood of business

[wentriken]

In our revenue cube, we need to model the likelyhood of business. When results go to the Summary P&L tab, forecasted business is "discounted" and aggregated. For example, customer A is forecasted to do $4m with 50% likelihood, add that to customer B with $5m @ 100% and the total is $7m discounted.

Currently we have this done in our measures list:

Revenue
Labor cost
Other costs
...
Probability of business
Discounted Revenue (calculated)
Discounted Labor cost (calculated)
Discounted Other costs (calculated)
... (calculated)


Is this the right way to accomplish this, or is there a more elegant approach?

[jeffowentn]

Just to clarify...when you say "measures list" it makes me think BI.  Are you referring to something in BI or something in Cognos Planning's Analyst or Contributor?

To answer your question, your approach is correct, regarding the expected results and the probability of them happening.  In your example, though, the $4M business at 50% likelihood suggests the other 50% probability is $0.  If that is correct, then you are good.

I really can't answer your question any further without knowing what the business needs are.  That will determine whether there is a better method to your approach.  I will add, though, that it looks like you just have one line for discounted labor costs and discounted other costs.  I suspect the business has fixed vs variable costs.  Assuming you use the same discount applied to the revenues, you are probably discounting your costs too much and not allowing for the fixed component of your costs to remain regardless of the revenue stream.

That's my two cents...hope that helps you out.

Jeff

[wentriken]

Thanks for your feedback.

Yes this is EP. These items are in a dimension a named Measures. Other dimensions in this cube are: customers, months, and budget versions.

I jumped in headfirst to design what we are now using for planning at a large enterprise, and am just trying to understand if any of the design decisions were poor and could be improved in the next version of the application.

Aside from the Go sample applications, are there a source of other free applications out there that I could compare with and look for best practices?

Thank you,
Will

[jeffowentn]

There are other models at the Cognos Innovation Center site...you will need to request the password from IBM Cognos to access the site and then download, unzip, and import the models into your environment.  There are functional-based models and industry-based models.

http://www-01.ibm.com/software/data/cognos/innovation-center/

Go to the Blueprints section...

Jeff

[wentriken]

Thank you, this is perfect. I will contact our account rep and get access to these.