Decision Matrix Generator

A coin flip is the right tool when two options are equally good. It is useless when they are good in different ways — a job that pays more but commutes worse, a supplier that is cheaper but slower. That needs a weighted score, and weighted scores go wrong in three specific ways: unnormalised weights that quietly double-count, criteria like cost where a high score should count against you, and a winner nobody checks for robustness. This handles all three, and shows the working.

Options you are choosing between

One per line — jobs, houses, suppliers, holidays.

What matters, and how much

Weights are relative — 3 against 1 means three times as important. They are normalised to add up to 100%, and the share each one ends up with is shown below.

Score each one out of 10

Scores for each option against each criterion
Option	Costlower is better	Qualityhigher is better	Speedhigher is better	Risklower is better
Option A
Option B
Option C

Score the options, then press the button.

What this generator does

Adds up your own reasoning rather than picking for you. The rankers and pickers on this site choose between options you cannot separate; this is for options you can separate but on several axes at once. It normalises the weights so a criterion scored out of 10 does not silently outrank one scored out of 5, inverts the criteria where less is better, and then tests how far a weight would have to move before the answer changes.

How to use this tool

List the options you are choosing between, one per line.
Set the criteria that matter, how important each is, and whether higher or lower is better.
Score every option out of 10 against each criterion.
Press score, then read the totals and the sensitivity note.

Understanding the controls

Options: One per line. Two or more, up to twenty. These are the things you are choosing between, not the reasons.
Criteria and weights: Weights are relative — 3 against 1 means three times as important. They are normalised to add up to 100%, and the resulting share is shown so nobody has to trust that 'importance 3' meant what they thought.
Higher or lower is better: The setting that stops the most expensive option winning on cost. A lower-is-better criterion is inverted before scoring, so a cost scored 9 out of 10 counts as a 1.
Scores: Each option against each criterion, out of 10. Score what you actually think rather than trying to engineer a winner — the point is to find out what your weights imply.

Worked examples

Salary 3, commute 1, interest 1: Weights normalise to 60%, 20% and 20%.
A commute scored 9 on a lower-is-better criterion: Counts as 1, not 9.
Two options scoring identically: Both shown as rank 1, with a note suggesting a criterion that tells them apart.
A margin under 0.25: Flagged as close enough that one weight change could flip it.

Common use cases

Choosing between job offers, houses, suppliers or holidays
Making a group decision where everybody weights things differently
Writing down why a choice was made, for somebody who will ask later
Finding out whether a close call is actually close
Turning a circular argument into a disagreement about weights, which is easier to settle

How this generator works

Weights are divided by their total so they sum to exactly 1. Lower-is-better criteria are inverted onto the same 0–10 scale. Each option's total is the sum of its scores times the normalised weights, so every total lands between 0 and 10 and every contribution is shown. Options that tie share a rank rather than being separated silently. The sensitivity figure is computed by re-running the whole matrix at a range of weights for the top criterion and reporting the first weight at which the leader changes — or reporting that it never does.

Randomness and fairness

Nothing here is random. The same options, weights and scores always give the same result, which is the point — a decision you can reproduce is a decision you can defend.

For how randomness is produced across the whole site, see how Generate Random works.

Assumptions this tool makes

Every criterion is scored on the same 0–10 scale, so the weights are the only thing controlling their relative influence.

Limitations and good to know

It adds up your numbers. If the scores are wrong the answer is wrong, and the tool has no way to know.
A weighted sum assumes the criteria trade off against each other. Where one is a hard requirement rather than a preference, filter on it first and score what is left.
It does not handle uncertainty in the scores themselves, only in the weights.
Nothing is saved between visits — copy the matrix before you close the page.

Common mistakes

Leaving cost as higher-is-better: Set it to lower-is-better. Otherwise the most expensive option gets the best score on it, which is the single most common error in a hand-built matrix.
Adjusting the weights until your preferred option wins: That is a way of discovering what you already wanted, which is genuinely useful information — but decide which you are doing.
Treating a 0.1 margin as a decision: The page flags small margins for a reason. If one weight change flips it, the honest conclusion is that the options are equivalent.

Practical tips

Set the weights before you score anything, so the scoring is not working backwards from a preferred answer.
In a group, have everybody weight independently and compare the weights rather than the conclusions.
If the winner surprises you, that is the most useful output the tool produces — work out which weight is wrong.

Privacy and your data

Your options, criteria and scores stay in your browser. Nothing is uploaded, nothing is stored between visits, and none of it is written into the page address — which matters when the options are job offers or suppliers. Analytics records that the tool ran and how many options were compared, never any text.

Frequently asked questions

Does this tell me the right answer?: No. It adds up the numbers you supplied using the weights you set. It makes your reasoning explicit and arithmetically consistent, which is a different and more honest thing than being right.
Why normalise the weights?: Because raw importance numbers do not add to anything meaningful. Normalising turns 'importance 3, 1, 1' into 60%, 20% and 20%, which is what those numbers actually meant, and shows it.
What does the sensitivity note mean?: It is the answer to 'how much would I have to change my mind for this to change?'. If the leader never changes however the top criterion is weighted, the decision is robust. If it flips at a small change, it is not.
What if two options tie?: They share the top rank and the page says so. A tie is a real result — it usually means you are missing a criterion that actually matters to you.