Evaluation Scorecard

An evaluation scorecard, sometimes called a decision matrix, is a table that connects the options under comparison, the mandatory conditions each must pass, the preference criteria on which eligible options are scored, the weights, the evidence behind every entry, and the resulting recommendation. Its purpose is to make a selection inspectable by someone who was not in the room. It supports the decision; the decision owner still makes it.

What a usable scorecard contains

Each option is defined precisely: product version, service tier, deployment, configuration, integration plan, and support scope, with the assessment date and workload. Gates are recorded as pass, fail, or unverified. Preference criteria show the raw measurement, its unit, the fixed anchors that turn it into a 0–100 score, and the weight. Every cell points to evidence with its source, date, reviewer, and status. A sensitivity section shows which changes in weights or inputs would change the order, and the recommendation states the conditions under which it holds.

For an invoice-evidence lookup with three fictional options, option A might score 0.50 × 80 + 0.30 × 40 + 0.20 × 60 = 64 and option B 0.50 × 60 + 0.30 × 80 + 0.20 × 80 = 70, while option C reaches 93 but fails the branch-isolation gate and is excluded. The scorecard reports B as leading among eligible options, C as excluded with the reason, and the weight shift at which A would overtake B.

What the total cannot say

A six-point lead is a statement about the agreed preferences, not a six-percent business improvement or a probability of success. The UK multi-criteria analysis manual treats the weighted total as the output of a linear additive model whose scales and weights were chosen by the assessors; change the anchors or weights and the total changes with them. NASA’s decision-analysis guidance adds that recommendations should assess whether reducing uncertainty could credibly change the ranking, and that recommending a lower-scoring option is a sign that the scoring did not capture the risks that mattered.

A scorecard therefore keeps its rejection rules, evidence gaps, dissent, and revisit conditions visible. A table of product names and unexplained stars is not a scorecard in this sense, because nobody can check what a star meant.

References: UK Government: Multi-criteria analysis manual, NASA: Decision analysis. Figures here are illustrative.