Decision matrix: weights the person deciding can move

Options scored against weighted criteria, with weights the person deciding can change.

Live example

The example: Three offices for Harkness & Vane Structural Engineers

Six criteria, live weights, one recommendation

Three offices for Harkness & Vane Structural Engineers

Open the exampleThe names and numbers are made up.

What to notice

  • It opens on a result: The Malthouse, 72.0 out of 100, 6.0 points ahead of Copper Row.
  • Every score has its evidence beside it, such as $318,000 a year next to a 2 out of 5 on cost.
  • Push Yearly cost to 9 and Eastfield Park takes first place by 0.9 of a point, as the section on flips says it will.

The prompt that starts one like it

Make a decision matrix as a single HTML page for [who decides]. We're choosing between [three or four options]. Score each from 1 to 5 on [criteria, with the fact behind every score], starting from these weights: [weights]. Give each weight a slider, and update the totals, the ranking and a sentence naming the leader as I drag. Add a reset button, which changes would flip the result, and how I scored. It must work on a phone. Then publish it on sharable.link with a password.

A weighted matrix has a known weakness: whoever sets the weights can make any option win. This one hands the weights to its readers. Two people choosing an office drag six sliders and watch the ranking move. It's an invented firm, so bring your own options and evidence.

Who it's for

Whoever did the legwork on a choice between three or four options and needs someone else to make the call: an operations manager with an office shortlist, a team lead with three vendor quotes. Send it a few days before the decision is due.

It suits arguments about priorities. If the disagreement is over a fact, such as the rent, settle that first. If one criterion is a deal-breaker, drop the options that fail it before scoring, because a weighted total averages a fatal flaw away. The page forgets the sliders on reload, so ask the deciders where they left them.

What a good one contains

  • A result on load. The sample opens on The Malthouse at 72.0 out of 100, 6.0 points clear of Copper Row. The recommendation is there before anyone touches a slider.
  • Few criteria. Six rows here. With fifteen, each weight is too small to change the outcome.
  • The fact behind every score. A 2 out of 5 on cost sits beside $318,000 a year. A score without evidence is an opinion with a number on it.
  • Weights shown as shares. Each slider runs from 0 to 10 and shows its share of the decision, so cost at 6 reads as 30%.
  • Points in every cell. Eastfield Park's low rent is worth 30.0 points and its long commute 5.0, which shows where each total comes from.
  • What flips the ranking. The sample lists the moves that change the winner: cost up to 9, or commute down to 2. Deciders see what they'd have to believe to choose differently.
  • How you scored. Cost and commute follow a stated rule, a point for every $40,000 or five minutes. The other four are labeled as judgment.

Score first, weigh second

Fix the scores before you look at any total. Adjust one after seeing who is ahead, and the matrix becomes a costume for the answer you wanted. Write down a rule wherever a criterion has a number behind it, and call the rest your read.

Then try to break your own result. Set every weight equal. Push each slider to both ends. In the sample no single slider puts Copper Row first, and it says so. If one small nudge changes your winner, say so. With commute at 3, the sample shows two offices level on 68.9.

How to send it

With a password. It sets real rents and lease terms beside your opinion of each option, and the landlords you turn down shouldn't see that. Two or three deciders can share one password.

Every page you publish gets one link that opens in any browser. You choose who can open it: anyone with the link, anyone with the password, or only the people and company domains you name, who sign in with Google or Microsoft. Change the page later and the link stays the same.

Make yours in Claude

Add the connector once, then publish from any Claude conversation and get a link to send.

Add Claude Connector

Already have the file? Upload it here.

More in strategy and finance

  • Business plan: A one-page plan an owner sends to a bank, a landlord or a partner ahead of the full version.
  • SWOT analysis: Strengths, weaknesses, opportunities and threats for one organization, each backed by evidence.
  • Gantt chart: One project's tasks as bars across weeks, with owners, milestones and a line for today.
  • Executive summary: One screen that tells executives what a longer report found and what they're being asked to decide.
  • Timeline: Dated events in order, for a history people will enjoy or a project they have to follow.
  • Org chart: Who reports to whom in a small company, with teams, open roles and what each person looks after.
  • All use cases

Keep reading