Build an Evidence-Linked Decision Matrix and Test Its Weights

What can a weighted decision matrix tell you?

Decision Brief turns your supplied options, criteria, weights, scores and evidence notes into an offline comparison and evidence ledger. The individual package is USD 4.99. Building it locally uses Python 3.10 or newer and no paid API. The result reflects your assumptions; it does not verify sources, choose for you or establish an objective best option.

Define comparable criteria and record evidence

Provide a decision context, 2–8 named options and 2–8 criteria. For each option, supply a score from 0 to 5 and an evidence entry for every criterion. Use consistent scoring direction: for example, a higher delivery score should consistently mean easier delivery, not more effort for some options. Explain the scoring scale and the time or evidence behind it outside the numeric score.

Give each criterion a weight from 0 to 100; their total must be positive. A zero weight excludes that criterion from the weighted average. Mark evidence supported, unverified or conflicting. Supported requires both a note and an HTTP(S) source URL; a link does not establish that the source is reliable.

Build and reconcile the fictional market example

Install the complete package and run this command from its folder. Choose a new output directory; the builder refuses to overwrite an existing one. It creates index.html, input.json and the initial brief.md. Open index.html in a modern browser.

• The calculation is sum(score × weight) divided by sum(weights). For Coast: (4×45 + 3×35 + 4×20) / 100 = 3.65. All nine evidence entries in this example are unverified; the displayed ranking is practice data, not market research or a launch recommendation.

Explore the fictional market comparison

python3 scripts/build.py --input examples/market.json --output ./market-practice

Test a change before relying on the ranking

Change Demand from 45 to 20, keeping Readiness at 35 and Delivery at 20. The total becomes 75; Coast shows 3.53, North 3.93 and Central 3.53 after rounding to two decimals. This is a sensitivity check, not proof that the revised priorities are correct. Compare decisions under plausible weights and investigate close scores and missing evidence.

Review evidence and preserve the edited comparison

• Use Edit for each option and criterion. Try marking an entry supported without a source: the editor should reject it. Review the source itself and record conflicting or incomplete evidence honestly. The supplied numeric score does not change automatically when its evidence status changes.
• Download comparison JSON before closing the page. The generated brief.md and input.json represent the original build, not later browser edits. To regenerate from the edited snapshot, use that downloaded JSON as --input and choose a fresh --output directory; review the rebuilt files.
• Document assumptions, unresolved evidence, risks, the decision owner and a review date. Share only authorized sources. The offline comparison does not publish, contact suppliers or provide live collaboration.

Translate reviewed research into a capacity-aware roadmap

Turn the chosen scope into a client proposal

Inspect the current example output → — Compare the visible result with the exported files and checks described in this section.