By James D. McCabe
Now that you've gathered data on top prospects that might help you with your networking projects, and established candidate ratings and prioritization, the sixth step is winnowing the list of candidates.Evaluating networking vendors paves the way for the last step of deciding whether networking vendors, networking tools and network service providers for your project will be evaluated first.
You now have an evaluation chart that shows how each network design candidate performed against various criteria. Weights were developed for the set of criteria. Each candidate received a rating for each criterion, which were then multiplied by that criterion's weight to form weighted ratings. The weighted ratings were combined to form a total for each candidate. The result of the evaluation is a set of overall or summary ratings that combines the individual ratings for each candidate.
FIGURE 10.11 Rankings of the candidates after evaluation
Summary ratings are used to prioritize the set of candidates. For example, from our previous evaluation chart we would get the prioritization shown in Figure 10.11. This figure shows the ranking of each candidate, along with the relative difference in ratings between that candidate and the one next higher in rank (relative delta), and the difference in ratings between that candidate and the top candidate (total delta). Either delta can be used to determine whether a candidate should be dropped from the set, or whether a tie should be declared. If, for this example, we had decided that a candidate had to be within one point of the next higher-ranking candidate (about 5% of the summary rating for Candidate 2) in order to declare a tie, we could see from the relative deltas that no candidates are close enough to be tied, and that Candidate 2 is the clear winner.
With a prioritized set you can eliminate candidates with bad ratings, with the goal of achieving a single optimal candidate. Usually a single candidate is chosen when it is clearly superior, that is, when its ratings are significantly better than all other candidates. Typically, on a first run of the evaluation there will not be one clearly superior candidate; instead, there may be few or several, depending in part on how many candidates you start with.
When the evaluation process yields more than one superior candidate, another iteration of the evaluation process should be performed on this reduced set. Successive iterations of the evaluation process begin with candidate discussions and continue through data gathering, criteria refinement and ratings development, ratings and prioritization, and another modification of the candidate set.
An advantage of doing another iteration of the evaluation process is that you already have most of the work done: the data sets and evaluation criteria. The work done in each successive iteration improves upon that information and should result in clearer and better ratings and prioritizations. Eventually, the evaluations should result in a single superior candidate.
Experience shows that often one or two iterations of the evaluation process will result in a single winner. If you do two evaluations and still have more than one candidate left (or worse, several candidates left), then you should reevaluate your evaluation data and how you are performing the evaluations.
Evaluating vendors and service providers for networking projects
How to choose vendors, tools and service providers
Seeding the evaluation process for networking partners
Having conversations about prospective networking partners
Gathering data on prospective networking partners
Refining your criteria for prospective networking partners
Developing ratings for prospective networking partners
Modifying the list of prospective networking partners
Determining the order of evaluations for networking partners
Reproduced from Chapter ten of the book Network Analysis, Architecture, and Design by James D. McCabe. Copyright 2007, Morgan Kaufman Publishers, an imprint of Elsevier Science. Reproduced by permission of Elsevier, 30 Corporate Drive, Burlington, MA. Written permission from Elsevier is required for all other uses.
This was first published in August 2007