A decision matrix can help a team zero in on the best engineering software when all the demos look alike.
Senior Mechanical Designer
Edited by Paul Dvorak
Making a buying decision based only on demonstrations is risky business, especially when trying to find the best CAD program to shape particular products. Most CAD demos are impressive displays of computer technology and operator skill. Demo jocks often make lightening-quick selections as colorful models take shape. And the last demo always seems to be the most impressive one.
A simple decision matrix, however, can help a team decide what's important by cutting through the mental clutter that builds up after several demos. Although the example here deals with selecting a CAD program, the matrix is easily redefined to select a machine tool, a finite-element program, or a new car.
A first step is to assemble a team of judges. In our case, a company manager invited three designers to participate in the project. Three is a manageable number and, being odd, it prevents ties when something requires a numerical score. Team members might also come from other departments when they are affected by the purchase.
The team decides on the selection criteria. The accompanying matrix shows a few items we considered important. For example, speed to us meant how fast we could build representative parts used in several projects. In this case, test parts were several hydraulic manifolds. Frequently used parts ensure that scores are not skewed by techniques needed to model unusual components.
We judged ease-of-use by the number of selections and menu picks a function required. The cost factor is straightforward. Stability was a requirement because our existing CAD system was prone to crashing. Most systems have tamed this problem now.
Associativity is a feature that updates a drawing when an engineer changes its model. An associative BOM (bill of material) automatically updates a BOM that appears in other documents when it changes on the associated drawing. Both save communication errors and updating omissions. You can see that the list could go on. For example, it might also include scores for training and reseller support.
The team also decided on the relative importance of each criteria. We decided, for example, that modeling assemblies is most important. One look at the accompanying image of an adhesive pump shows why. Many machine orders are low volume, one to three units, so there is no assembly line on which to work the bugs out. Machines must assemble correctly the first time.
Speed and a short return on the investment are also important, thereby scoring 9s in the Factor column. Your weightings or factors may be different.
Each person in the team can fill in a matrix after the top, left, and right side of the grid are complete. This is a good time for demonstrations. However, the tactic we chose was to borrow each of the candidate CAD systems for 30 days. This let us judge ease-of-use, how long a learning curve to anticipate, and when we might expect a return on investment. Only one software developer refused the 30 day loan, so we visited its facility.
The total in each column is the sum of the category's score and its factor for each row. Take CAD A, for instance. The speed rating of 5 and factor of 9 ( 5 × 9 = 45) contributes 45 points to the total of 487. The product with the highest score "wins." In this case it was CAD B.
The decision matrix lists criteria on the left and candidate products across the top. A weighting or importance factor appears on the right. Total scores are the sums of each criteria rating times its factor.