Pedagogical Product Line
Overview
Business Case
Scope
Requirements
Concept of Operations
Architecture
Unit Test Plans
Production Plans
System Test Plans
Brickles Product
Pong Product
Bowling Product
Bibliography
Glossary
Misc Documents

Arcade Game Maker Memo 04-03

To: For the record
From: AGM Product Line Manager
CC: Vice President for Product Development
Date: May 22, 2004
Re: Business Model Update and Decision Support

Since we developed our original business case, a seminal paper1 was published on a business model that includes the equation below. We used this model to reanalyze our business case and quantify some difficult business decisions.

In this memo, we analyze and apply the equation to illustrate how it supports a fundamental decision made by our product line organization.

Equation's Fixed Costs

Corg is the cost of readying the organization for the product line strategy. It appears that our original estimate of $30,000 was accurate.

Ccab is the cost of the core assets. Because the products are being produced in three groups of three products, any core assets that apply to the second and third sets do not have to be available until that increment of the product line is ready to be produced. Therefore, we have two choices: we can produce all of the core assets at once up front, or we can produce them incrementally, just in time. In this memo's Development Decision section, we explain how this model helped us make that decision.

Equation's Per Product Costs

Cunique is the cost of constructing the unique portion of each product. In Memo 04-02, we used a new, unpublished metric to estimate a homogeneity value of .8 for our products2. Based on our original estimates of .3, .6, and 1.0 for our three product groups, we can now estimate savings of $1,250,000 as opposed to $1,000,000. This estimate is based on a raw homogeneity value that has not been calibrated to our development environment.

Creuse is a function of the quality of the core assets: it is the cost of reusing the core assets in the context of the product. Because the cost of building a reusable asset (Creuse) is often estimated at 1.5 times the cost of building for a specific use (Ccab), Creuse decreases to some extent as Ccab increases. Further investigation of this relationship would allow us to determine what level of development for reuse produces the best return.

Development Decision

As stated above, we could build all of the core assets at one time before building any products, or we could build only those assets required for the next increment.

If we build all assets at once, some remain unused until the specific increment is built. Therefore, the value of Ccab must include an opportunity cost, which is based on the value of having resources used to create the core assets available for other work during that time (see figure). In other words, are some developers doing work that will have to wait if they are diverted to core asset development, and if so, what is the value of that work?

The model allows us to select the implementation of Ccab, giving us the opportunity to represent the two approaches and compare the costs. We can assume that it will cost the same to implement the assets at either point. Thus, if the opportunity costs are greater than zero, the choice is obvious–build just in time. If opportunity costs are zero, it doesn't matter.

Our opportunity costs are greater than zero. We don't have enough developers to construct all the core assets up front. But no more developers will be available later, so we will build assets continuously rather than just before the next increment.


Cost of Inventory


1 Bockle, G.; Clements, P.; McGregor, J. D.; Muthig, D.; & Schmid, K. "Calculating ROI for Software Product Lines." IEEE Software 21, 3 (May/June 2004): 23-31.
2 The accuracy of this estimate depends on how similar the use cases are in terms of the amount of effort required to implement them (i.e., as the differences in use cases increase, the accuracy of the estimate decreases).