So now you have your needs analysis done. Put it in a drawer for a couple of days and do something else, and then pull it out again and go through it. Be harsh with your evaluations to determine whether the "must haves" really are must haves.

I've seen needs analysis that were for relatively simple projects that put them into the NASA league, and believe you me - you don't want to go there. You end up with a very very narrow scope of potential candidates and end up turning blind eyes to better and more flexible products. I had one where the chief decision maker was very rigid and it ended up costing the organization $250,000 to purchase the product, whereas a little more flexibility could have gotten a free Open Source product.

Ok - back to the needs analysis. One potential use for a needs analysis is to be able to quantify how good a fit a product has to your requirements. The way to do this is to assign a points value to each criteria. Obviously, if a product cannot meet your 'must haves' then it is eliminated, but it becomes difficult sometimes to distinguish between the other candidates. So get your calculators out..

Assign a points value of 100 for each 'must have', 50 for each 'very important', 20 for each nice to have and 5 for each 'not important' and then grade the product. For each requirement, estimate how well and completely the product meets the requirement.

You will note the inconsistency here - I said if a product doesn't meet your 'must haves' then you eliminate it, and then go on to describe scoring it. This is because there are many ways to meet a requirement, and some are significantly easier to deal with than others. If my requirement was to ship electronic format product between a variety of geographic locations, then I could do this by copying diskettes or CD's and shipping them, or by electronic communications. You have to decide what that means to your organization.

At the end of this exercise you have a spreadsheet that allows you to compare like products. The one with the highest score is most likely your best candidate, but a note of caution here. The real use behind this is to allow you to COMPARE the products in a quantifiable way - but not to automate the decision process. Please, please, please make an informed decision, not just a mathematical one.

So far, the process we have described can apply to any product, both open source and commercial. With open source products, it is usually easier to get an unbiased opinion of how well it meets needs and in our next article we will be looking at ways of getting those opinions.

About the Author
Lee is one of the principals at Spinnaker Systems which provides Web related services to the small business owner. Lee can be contacted at lee@spinnakersystems.com and is a regular contributor to the Spinnaker Blog