Weighted Decision Matrix
Build a small comparison from criteria you define. Each raw score is normalized against your fixed minimum and maximum; a lower-is-better criterion reverses that scale. The tool shows each criterion contribution, ranked totals and changes under explicitly varied weights.
Key features
- Fixed scales keep one alternative from changing another alternative normalization
- Benefit or cost direction per criterion
- Normalized weights, per-criterion contributions and tied ranks
- Eleven weight scenarios plus a custom percentage
- Actual CSV and JSON downloads with no automatic storage
How to use
- Load the example or enter alternative and criterion names.
- Set scale bounds, preferred direction and nonnegative weights.
- Enter every score and calculate.
- Select a criterion and inspect scenario leaders and totals.
- Download the result; edits remove it until recalculation.
Use cases
- Compare workshop materials using your own preferences
- Inspect how setup-time weight changes a study-tool comparison
- Discuss assumptions behind a team planning choice
Frequently asked questions
How is the total calculated?
Benefit utility is(score−minimum)/(maximum−minimum); cost utility is1 minus that value. Each utility is multiplied by normalized weight and100, then summed. This is a linear additive model; it does not infer preferences or interactions.
Does normalization depend on other alternatives?
No. Bounds are supplied by you and stay fixed. Removing an alternative does not rescale remaining scores. Every score must lie inside its criterion bounds.
What does sensitivity change?
One criterion receives the chosen percentage. Other criteria share the remainder in their original raw-weight ratios. If all other weights are zero, scenarios below100% are unavailable rather than assigning invented ratios. The10-point grid does not locate every crossover; the custom percentage tests a specific value.
How are ties and missing scores handled?
Totals differing by no more than1e−9 points share a competition rank. Display rounds to4 decimals; exports retain numeric precision. Missing scores, including on zero-weight criteria, are rejected.
What limits and export formats apply?
Use2–12 alternatives and1–12 criteria,80-character names, weights0–10,000 and bounds within±1,000,000. Decimal input accepts up to6 places, without scientific notation. CSV contains scores/contributions and criteria; JSON also records input and sensitivity. Formula-like CSV strings receive a protective apostrophe.
Does a first rank guarantee a correct decision?
No. Rankings reflect only your entered scores, linear assumptions and weights. They are not guarantees, financial recommendations, hiring assessments or automated decisions. Revisit the criteria and evidence yourself.
Privacy
Inputs and results stay in page memory. No remote request, analytics payload or automatic storage is used by this tool. Downloads are created only when requested.
Comments & questions