01 / START HERE
A useful way to approach it.
A decision matrix makes your priorities visible. It does not turn subjective judgments into objective facts. Use one when several options meet a basic requirement but involve different tradeoffs. Eliminate options that fail a non-negotiable constraint before assigning scores.
Choose a small set of criteria that do not count the same benefit twice. Define the scale before rating the alternatives. Add an evidence or uncertainty note to every score and test whether a modest change in your assumptions changes the ranking. The exercise is useful even when it shows that you need more information.
02 / THE PROCESS
Go from question to next step.
- 01
Separate constraints from preferences
List requirements an option must meet, such as a budget ceiling or required export format. A weighted total should not let an attractive but ineligible option overcome a mandatory requirement. Mark eligibility before comparing preferences.
- 02
Define criteria and weights
Write four or five distinct criteria and distribute 100 weight points. Have the decision owner explain the weights. Cost and affordability may measure the same thing; remove overlap so one preference does not receive extra influence.
- 03
Anchor the scoring scale
Use a scale such as 1 to 5 and define what the endpoints mean for each criterion. Record a source, test result or explicit judgment beside each rating. Leave an unknown unscored rather than silently assigning it the middle value.
- 04
Calculate a weighted result
For each scored criterion, multiply its weight by its score divided by the maximum score. Sum the contributions only when the same criteria have evidence for all eligible options. Preserve both the original ratings and the calculation.
- 05
Test the recommendation
Change the most uncertain score or an important weight and recompute. If the leading option changes easily, call the ranking fragile. Identify the single missing observation most likely to improve the decision, then make that your next step.
03 / YOUR STARTING POINT
A prompt you can make your own.
Replace the bracketed fields. Add only information you have permission to share. This is a reusable text prompt; it does not run a task from this page.
Build a business decision matrix. Decision: [choice]. Alternatives: [options]. Mandatory constraints: [pass/fail requirements]. Criteria and weights totaling 100: [list]. Scale anchors: [what 1 and 5 mean for each criterion]. Evidence: [source-backed ratings or observations]. Separate constraints from preferences. Do not invent ratings for missing evidence. Show weight × score / 5 for each contribution. Flag double-counted criteria. Run a sensitivity check using the most uncertain input and explain whether the recommendation changes. End with the next evidence-gathering step; I make the final decision.
04 / SEE THE SHAPE OF IT
Worked calculation: choosing an internal documentation tool
Fictional options and subjective example ratings. Scores show arithmetic, not tested product quality.
| Criterion / weight | Option A contribution | Option B contribution |
|---|---|---|
| Editing fit / 40 | 40 × 4 / 5 = 32 | 40 × 3 / 5 = 24 |
| Export fit / 35 | 35 × 3 / 5 = 21 | 35 × 4 / 5 = 28 |
| Setup effort / 25 | 25 × 3 / 5 = 15 | 25 × 4 / 5 = 20 |
| Total / 100 | 68 out of 100 | 72 out of 100 |
If Option A's editing rating changes from 4 to 5, its total becomes 76. That reversal means the editing assumption deserves a real test before treating Option B's original lead as a decision.
05 / KEEP YOUR JUDGMENT
What to watch before you use it.
Precision without evidence
A total like 72 is a calculation of judgments, not a measured probability of success. Keep the ratings and evidence next to it.
Unknowns scored as average
A middle rating hides missing information. Mark an unknown and obtain comparable evidence before computing a final ranking.
A matrix replacing judgment
A score cannot take responsibility for risk or approve spending. Use the matrix to explain a human decision.
06 / THE FINAL PASS
A little review goes a long way.
Your checks stay on this page. They do not approve or run a task.
07 / GOOD QUESTIONS
Before you begin.
Should every decision use a matrix?
No. It is most useful for explicit tradeoffs among eligible alternatives. A simple pass/fail check may be enough for a small or reversible choice.
Can I let AI set the weights?
Ask it to suggest questions or a starting structure. The decision owner should choose and approve the priorities; an assistant does not know your tradeoffs automatically.
What if the scores are close?
Treat that as a reason to check uncertain inputs, not to add decimal places. A small real test may be more useful than more elaborate scoring.
