Field guide
Choose criteria you can explain
A good comparison starts with clear meanings for the rows.
Start with a sentence
Imagine a fictional neighborhood club choosing its next Saturday meetup. Its options are an indoor discussion, a photo walk, and a postcard workshop. Before anyone scores them, the group writes: “We want a two-hour meetup that encourages conversation and is manageable for one volunteer host.” That sentence gives the comparison a purpose. Without it, one person may reward novelty while another quietly optimizes for the easiest afternoon. Both judgments can be reasonable; they answer different questions.
Separate requirements from preferences
Some conditions decide whether an option belongs on the board at all. If the group needs permission to use a space, an unavailable space is not rescued by an attractive total. NASA distinguishes mandatory criteria from qualities that improve an otherwise feasible option. Treat a hard requirement as a separate check. In our invented scenario, all three formats are assumed feasible; that assumption would need checking for a real event. NASA decision analysis guidance.
Give each row one job
“Convenience” is too broad if it means preparation for the host, travel for guests, and how easily a plan can change. Split it only where those are distinct concerns. Conversely, avoid counting the same benefit under several appealing names. The government’s methodological manual warns that duplicated effects can acquire extra influence in a total. Multi-criteria analysis manual, section 5.4.4.5.
For the club, “low preparation” can mean the host’s estimated preparation hours. “Weather resilience” can mean how much of the activity remains usable in rain. Those definitions tell a scorer what belongs in each row. They also tell a reader what has been left out.
Keep the observation beside the judgment
Write “assumed preparation: 90 minutes” separately from “low-preparation score: 4.” The first is an estimate about the option. The second expresses how that estimate fits the chosen scale. Neither becomes verified because it sits in a table. Add a short note such as “fictional estimate” or, for a real comparison, the source and date. When a fact is missing, name the gap rather than letting a tidy number conceal it.
Define the score before assigning it
Choose clear meanings for zero and five, then describe useful steps between them. For preparation, a five might mean at most one hour, while a four means more than one and up to two hours. Keep higher scores favorable in every row. A weighted sum assumes that equal score changes represent equal preference changes within a criterion. If that feels wrong, revise the scale or explain why the number is only a rough aid.
Weight the range you actually mean
An important-sounding label is not enough to justify a large weight. Compare the improvement from your defined zero to your defined five. UK guidance uses this range-based idea in swing weighting and warns against weighting criterion names alone. Government Analysis Function guide.
Finish by reading every row aloud: what it means, what supports each score, and why its full improvement matters. If someone disagrees, keep that disagreement visible. The useful result is a comparison whose judgments can be inspected and revised. A number is the beginning of that conversation.