Criteria before solutions — qualitative and quantitative
◈ 6 cardsFix the criteria from the stakeholder objectives before any alternative is scored; label each qualitative or quantitative; both kinds belong in the matrix.
Criteria come first
The second and third lines of R-psp-3: criteria named, each traceable to a stakeholder objective, and qualitative and quantitative criteria both present. The order of work matters more than it looks: the criteria are fixed before the alternatives are scored — ideally before they are generated — because a criterion written after a favourite has emerged is written to favour it. Module 4 converted Lakeshore's stakeholder objectives into criteria; stage 3 picks them up, adds a measure to each, and labels them.
Worked example — Lakeshore's five criteria
| Criterion | Traces to | Kind | How it is scored |
|---|---|---|---|
| Annual cost | Priya (cash); the money constraint | quantitative | dollars per year: repairs, or loan payment plus repairs, or lease, or margin lost |
| On-time reliability | the key accounts; Priya (reliable production) | quantitative proxy | expected on-time %, judged from the press's age and the maintenance terms |
| Cash-flow impact | Priya (preserve cash); the bank (debt capacity) | quantitative | up-front outlay and new debt |
| Staff impact | the two operators; the other four staff | qualitative | jobs kept or lost; a judgement, justified |
| Flexibility | Priya — volumes may change | qualitative | ease of exit or upgrade; lock-in |
Two things to see. Every criterion has a stakeholder beside it — that is the traceability line — and the two shared by several stakeholders (cost, reliability) will carry the most weight in the next lesson. And the kind is labelled: three quantitative, two qualitative. Quantitative means the score comes from a number the case supplies or a measure that could be observed (dollars, a percentage, weeks); qualitative means the score is a judgement — and a judgement is scored like anything else, with its reason written in the cell. The error is to drop staff impact because it has no dollar figure; the second error is to invent one ("redundancy costs $40,000 of morale") — false precision is worse than an honest judgement.
The retro-fit trap
Suppose the lease is looking good and someone adds a sixth criterion, vendor relationship, on which the lease scores 5 and everything else 2. It traces to no stakeholder objective written in stage 1; it appeared after a favourite emerged; and it fits one alternative. That is retro-fitting — a bias, and the second most common way a matrix is rigged (the first is the weights). Any criterion added after scoring has begun must trace to a stage-1 objective or be struck.
A related check: a criterion that only one alternative can satisfy — includes maintenance, say, which only the lease does — is an alternative's feature dressed as a criterion. The criterion is reliability or annual cost; included maintenance is a reason for a score in those rows.
Derive the clinic's criteria
From module 4's objectives: no-show reduction (dentists; quantitative — expected fall in the 12 %); annual cost (dentists; quantitative — subscription dollars); scheduling accuracy (Tomas; quantitative proxy — double-bookings per month, expected); ease of use / training burden (receptionists; qualitative); patient convenience (patients; qualitative — a judgement about who will use it). Three and two; every one traces to a name.
Classify cold
For a supplier decision (module 11's Kestrel case): gross margin on their goods — quantitative; on-time % — quantitative; defect rate — quantitative; sustainability audit grade — qualitative (a letter, but a judgement behind it); strategic fit — customers ask for the line by name — qualitative. Both kinds, and every one belongs in the matrix.