Evaluation Matrices

An evaluation matrix is a structured comparison of options against weighted criteria. The matrix does not make the decision — it makes the reasoning visible. When stakeholders disagree, the disagreement can be traced to a specific criterion weight or a specific score. "You scored latency higher than I did because your team is consumer-facing and mine is batch processing." The matrix transforms subjective debate into traceable disagreement.

1. Define Criteria from Requirements Extract evaluation criteria directly from the requirements. Performance, scalability, operational complexity, team expertise, total cost of ownership, vendor risk. Each criterion maps to a requirement. A criterion that does not trace to a requirement is a preference, and preferences do not belong in the evaluation matrix.
2. Weight Criteria by Stakeholder Priority Assign weights before scoring options. If performance is twice as important as operational complexity for this decision, weight it accordingly. The weights are set by the stakeholders who will live with the consequences. Setting weights after scoring is reverse-engineering the conclusion — it looks rigorous but it is rationalization.
3. Score with Evidence Score each option against each criterion using evidence: benchmarks, case studies, team self-assessment, vendor documentation. A score without evidence is an opinion. Document the evidence alongside the score so future readers can evaluate whether the evidence is still valid.