Decision Making

A decision matrix provides a systematic approach to evaluation and selection. The evaluation is made against stakeholder requirements, goals, and benefit to the organization.

The Pugh (Stuart) matrix is one of the simplest techniques in this category. The approach supports fast qualitative comparisons by comparing multiple options to one case or a baseline. It is an effective framework for discussion on reducing vendors to a short list, narrowing potential solutions, or concept selection. It is a good fit early in a process; enough detail to foster a quality conversation, but not too much rigor to get bogged down. It provides evidence of a structured approach to decision making on a project by the team, and process documentation (traceability) for any questions that arise during final approval. Every member is involved, and team ownership is fostered.

In its simplest form, the letter S is entered in a cell where there is little or no difference, a plus or minus (+/-) where there is a meaningful advantage or disadvantage. The plus/minus model can be replaced with +/-1, 2, 3 for more granularity in each category. A procurement example: if all vendors appear to have the required technical skills this is noted with S (same). Where there is differentiation in a proposal, a plus or minus can be entered. If a bidder has not included information in a category, a blank entry is suggested. The categories are summed. The decision sequence is: Select by the most plus signs; discard the most negative signs; weight the final choice by number of same (S). Net scoring can be used with + and – cancelling each other; 0 replacing S with no contribution to result. Net scoring cannot be relied on in all cases due to the subjective nature of this model, especially if a simple un-weighted examination is used.

Grouping individual categories is helpful and can help distinguish importance; for example, organizing by each stage in the project life-cycle (design, implementation, support, transaction, etc.) Separating cost into a second matrix is a good practice; this helps to focus the discussion on the best selection. It’s important to identify if certain categories are highly sensitive to one criteria, and if that criteria important.

Greater refinement can be established by starting with a weighed ranking process (e.g., pair-wise) and using these values as the entries, plus or minus, for each cell.