The Weighted Decision Matrix, Explained

After reading this you will be able to score a handful of options against weighted criteria, compute an honest ranking by hand, and check whether your winner survives a small change in the weights or is just a coin toss dressed in numbers.

What a weighted decision matrix is

A weighted decision matrix turns a fuzzy choice into arithmetic you can inspect. You list the things that matter (the criteria), give each one a weight, then score every option on every criterion. The tool multiplies scores by weights and adds them up. The option with the highest total wins, and you can read off exactly which criterion carried it.

Here is the hook. Suppose you are choosing an apartment and you score three of them on rent, commute, size and neighborhood. Downtown scores badly on rent (4 out of 10) but wins on commute (9). The suburb wins on space (9) but the commute is brutal (3). Which one wins depends entirely on how much you weight commute against rent. The matrix makes that dependence visible instead of leaving it in your gut.

This method is often called Kepner-Tregoe scoring after the two authors who popularised it for management decisions in the 1960s. The math is plain multiplication and addition. The value is discipline: you have to write the criteria down.

When to use it, and when not to

Use a weighted matrix when you have a small set of options (roughly two to eight), several criteria that genuinely pull in different directions, and you can score each option on a consistent scale. Apartments, job offers, laptops, vendors and venue choices all fit.

Do not use it when one criterion is a hard requirement rather than a trade-off. If a laptop under 16 GB of RAM is unusable for your work, do not give RAM a weight, filter those laptops out first. A matrix that averages a dealbreaker against nice-to-haves will happily recommend something you cannot actually use.

The formula and the intuition

For an option with scores s_1, s_2, \ldots, s_n against criteria with weights w_1, w_2, \ldots, w_n, the weighted total is a dot product:

Here n is the number of criteria, w_i is the weight of criterion i, and s_i is that option's score on criterion i. Only the ratios of the weights matter, so weights of 5, 4, 3, 2 give exactly the same ranking as 10, 8, 6, 4. Doubling every weight doubles every total and changes no order.

The three normalization modes rescale the inputs before this sum:

Raw weighted sum

Uses your weights and scores as typed. Totals are easy to trace back to the numbers you entered.

Weights normalized to 100%

Divides each weight by the sum of weights, so they add to 1. The ranking is identical to raw; only the scale of the totals changes, which makes each criterion's percentage share readable.

Scores rescaled 0 to 10 per criterion

Stretches each criterion's scores so the worst option sits at 0 and the best at 10. This stops a criterion you scored 0 to 100 from silently outvoting one you scored 1 to 5.

A worked example with the demo data

Reading the contributions and the ranking

The tool breaks each total into per-criterion contributions, which is where the story lives. For the suburb house, rent contributes 40 of its 91, or about 44%. For the midtown flat, no single criterion dominates: rent gives 30, commute 28, size 18, neighborhood 14. The suburb wins because it maxes out on the two heaviest criteria you scored it well on, not because it is broadly good.

Switch to weights-normalized mode and the numbers become shares. Rent's weight becomes 5/14 \approx 0.357, so the suburb's rent contribution is 0.357 × 8 ≈ 2.857 and its total is 91/14 ≈ 6.5 on a 0 to 10 scale. The order does not move, because dividing every total by 14 preserves rank. Use this mode when you want to say "rent drove 44% of the score" rather than quote a raw 40.

The sensitivity check, the honest part

A ranking that flips when you nudge a weight by a hair is not a decision, it is noise. The tool asks, for each criterion, the smallest weight change that would crown a different winner. Do this by hand for Commute in the demo data.

Let the Commute weight be w instead of 4, keeping the others at 5, 3, 2. The suburb and midtown totals become:

1. Suburb: 40 + 3w + 27 + 12 = 79 + 3w.
2. Midtown: 30 + 7w + 18 + 14 = 62 + 7w.

Midtown overtakes the suburb when 62 + 7w > 79 + 3w, that is 4w > 17, so w \gt 4.25. Raising the Commute weight from 4 to just above 4.25 flips the winner. That is a 6% nudge on one weight. The suburb's victory is fragile.

Contrast that with a criterion where no realistic change flips anything. If sweeping a weight from 0 to five times its value never changes the top spot, the winner is robust and you can stop agonising over the exact number.

Common mistakes

Four errors account for most bad matrices.

1. Mismatched score scales. Scoring price 0 to 100 and warranty 1 to 5 lets price dominate no matter what weight you set. Use rescaled mode, or score everything on one range. If the ranking changes between raw and rescaled, your scales were deciding, not your preferences.
2. Direction confusion. Decide once that higher always means better. If rent is a cost, score cheap as 9 and expensive as 3, not the price in dollars.
3. Padding the criteria list. Adding five low-weight criteria dilutes the two that matter. A weight of 1 among weights of 5 contributes at most one seventeenth of the differences. If a criterion cannot change the ranking, drop it.
4. Ignoring a fragile winner. A one-point lead that flips at a 6% weight change (as in the demo) is a tie. Treat the top two as equally good and decide on something the matrix did not capture.

Related tools

Once you have chosen the option, the work of delivering it begins. To estimate when a project will finish from your real throughput, use the Monte Carlo Project Forecast. To find which tasks set your deadline, see the Critical Path (CPM/PERT) Calculator. To understand why loading more work in progress slows delivery, try the WIP Limits & Little's Law Simulator. And to pack chosen items into boxes or trucks, use the Box & Bin Packing Calculator.

Related tools

Box & Bin Packing Calculator

How many boxes, bins or trucks? First-fit-decreasing packing by capacity or 3D size.

Critical Path (CPM/PERT) Calculator

Which tasks actually set the deadline — schedule, slack and critical path from a task list.

WIP Limits & Little’s Law Simulator

Watch a kanban board run: raise the WIP limit and see cycle time explode while throughput stalls.

Round-Robin Schedule Generator

Everyone plays everyone: fixtures by the circle method, byes included.

Seating Chart Optimizer

Assign guests to tables around keep-together and keep-apart rules.

Secret Santa Generator

Draw gift-exchange pairs with exclusions, no self-draws guaranteed.