Which & Why package doesn't just organize your data--it tells you what to do
Which & Why decision-support software for Microsoft Windows 3.x not only helps you organize and analyze your options statistically, but also makes its own recommendation. The software, which runs on a Windows PC with at least 13M free on the hard drive, or on Novell NetWare or Banyan Systems Vines networks, breaks up decisions into bite-size factors, weighted for importance.
Which & Why decision-support software for Microsoft Windows
3.x not only helps you organize and analyze your options statistically, but also makes its
own recommendation.
The software, which runs on a Windows PC with at least 13M free on the hard drive, or
on Novell NetWare or Banyan Systems Vines networks, breaks up decisions into bite-size
factors, weighted for importance.
Let's say you're shopping for a new financial system. You have five overall criteria:
compatibility with your legacy system, compliance with government standards, ability to
handle 500 users, support for electronic data interchange and good report-writing
features.
The Which & Why interface lets you establish up to 10 such main factors. Each can
be broken down into four levels with 10 more sub-factors per level, giving you hundreds of
things to compare.
For your imaginary financial system, create sub-factors under the compatibility
heading, titling them "Digital Unix" and "Windows clients." If the
ability to run on legacy Digital Equipment Corp. machines is essential, weight that at 10.
If you'd like Windows client support, but it's not essential, weight that factor at 4 or
5.
Down one more level, put a sub-sub-factor under Windows clients for something like
"X Window capability." And continue on, to whatever level of detail you want.
You've built a hierarchical decision model for quantifying all the potential factors. Now
comes the fun part.
Ask Which & Why for its recommendation and then watch the what-if scenarios as you
change your factors or their weighting.
You'll be surprised at the results. The most expensive financial system might turn out
to be your best choice because it contains the things you've weighted most heavily.
Conversely, you may discover it has too many expensive bells and whistles.
Which & Why looks at a choice from two perspectives. The first is a weighted
average: Do I like what this choice offers me, and how much do I like it? The second is a
matching index: To what degree does this choice contain what I'm looking for?
Combining the two perspectives in a proprietary pattern recognition technique shows
whether the option you like best also is consistent with what you're looking for.
Gut instinct it's not. This is meat-and-potatoes numerical analysis, and the fact that
the $349 package tries to remove emotions from the equation is what makes it so valuable.
It just might keep you from doing something silly, but only you can set the factors to
rate. It can't do everything for you.
Keep the manual on your lap, at least until you've made your first model. There are
on-line help files, but they contain a lot of product cheerleading, and the help viewer
locked up on me twice.
Viewing your data in many different ways is a wonderful concept that takes work. You'll
learn for yourself when a pairwise comparison graph is better than a chart of weighted
averages. Finding the best view from an almost infinite supply of charts takes trial and
error. And the best view likely will differ from project to project.
But Which & Why is worth learning to use. My vote for the coolest function goes to
how you discount various options to equalize them. This works best with prices--you see
what discount must apply to make the different choices equally desirable. For instance,
you could with confidence tell a vendor, Hey, knock $2,361 off that price and I'll take
it.
Which & Why's sample decision models show a snapshot of what can be done, but only
a few are complete, and only one looked really useful for system purchases.
When decisions are made by committee, Which & Why lets groups team up to build a
single model. Or members can work individually, then compare and debate their results. The
influence of each member can even be weighted to build a modified consensus.
If the boss really wants that server with the shiny blue cabinet, he can skew the
statistics. But everyone's going to know he did it.
Arlington Software Corp., Baie d'Urfe, Quebec; tel. 613-746-1140.