Field guide
Would one weight change your choice?
Use a small sensitivity check to see where your comparison depends on a judgment.
Treat the first ranking as a starting point
A comparison can look decisive because its inputs happen to favor one option. Before acting on the total, ask whether a reasonable change in one priority would change the leader. This is a sensitivity check: a way to see which assumptions deserve attention. It does not tell you how likely an outcome is. UK guidance describes changing inputs individually before examining combinations. Government Analysis Function guide.
Save a baseline
In our fictional meetup example, the weights are conversation 30%, low preparation 30%, weather resilience 25%, and novelty 15%. Scores run from zero to five. The indoor discussion scores 5, 4, 5, and 2. The photo walk scores 4, 5, 1, and 5. The postcard workshop scores 2, 3, 5, and 4.
Multiply each score by its weight and add the contributions. The discussion reaches 4.25 out of 5; the walk reaches 3.70; the workshop reaches 3.35. On a zero-to-100 display, those are 85, 74, and 67. The display is a rescaled preference total, not a success percentage. Save these inputs before trying alternatives.
Change one share consistently
Suppose the group wants to explore a lower emphasis on weather. Reduce weather’s share from 25% to 10%, while keeping the other criteria in their original proportions. The remaining share grows from 75% to 90%, so multiply each other weight by 90/75. The new weights are 36%, 36%, 10%, and 18%. Keeping other weights proportional is an established way to isolate a weight’s effect. Multi-criteria analysis manual, Figure 7.20.
At those weights, the discussion scores 4.10, the walk 4.24, and the workshop 3.02. The walk now leads. Write down the redistribution rule: moving all fifteen percentage points to novelty would answer a different question.
Find the boundary worth discussing
For this example, let w be weather’s share as a decimal. Keeping the other weights proportional gives the discussion a total of 4 + w and the walk a total of 4.6 − 3.6w. They tie when w is approximately 0.1304, or 13.04%. Below that share, the walk leads. Between it and 100%, the discussion leads; at 100%, the discussion and workshop tie.
That boundary makes the conversation concrete. Would the group reasonably put weather below 13%? If yes, the preferred format depends on a live disagreement. If every plausible view puts weather between 20% and 35%, this particular disagreement does not change the leader within that range.
Check the limits of the check
A changed priority is different from changed evidence. If an indoor fallback becomes available for the walk, revisit its weather score separately. Then consider combinations that could plausibly change together. NASA recommends documenting the assumptions, limitations, and uncertainties behind a comparison. NASA decision analysis guidance.
Finish with a bounded statement: “The discussion stays ahead across the weather weights we tested,” or “The lead changes near 13%, so we should agree on that tradeoff.” Keep the tested range and unchanged assumptions with the export. One stable result does not establish that all weights, scores, or omitted criteria are harmless.