Build the decision framework: criteria (the things that matter) each holding one or more sub-criteria (the specific, scorable aspects). Criterion weight and within-criterion sub-criterion weight combine into each sub-criterion's model weight — its share of the final result.

Score every option 1–10 on each sub-criterion where higher is always better. Add a short description to keep everyone scoring the same thing.

Leave cost out of the criteria. Money belongs on the Cost & Benefit tab, for the reasons set out in the note below. Judgements about cost — such as how much confidence you have in an estimate — are perfectly good criteria, because they are about risk rather than the amount.

Each criterion has an Included / Excluded switch. Switching one off drops it from the model and redistributes its weight across the rest, while keeping every score it holds — switch it back on and the scores return. Useful for testing how much a contested criterion is really carrying, without deleting anything.

Weights are a starting point — the Sensitivity tab shows whether changing them would actually change the decision.

Pick the person scoring, then score each option 1–10 on every sub-criterion (higher = better). Switch person and repeat. Each person's scores are kept separately and the group average drives the result.

As scores go in, a small hint under each option shows the group's current average and range, and the bar at the bottom tracks the leading option, the margin and how many disagreements could actually change the outcome — so you can steer the conversation live.

The ranked outcome from the group's average scores, with the decision margin between first and second. A wide margin is a robust decision; a narrow one is finely balanced.

The consensus matrix shows every option's average per sub-criterion, colour-coded by how much the group disagreed. Use it to spot where opinion is split before you look at whether that split matters (Sensitivity tab).

This is the point of the tool. It separates disagreements that change the decision from disagreements you can safely park. If two people scored a sub-criterion 1 and 10 but the winner is the same whether the group lands on 1 or on 10, there is nothing to resolve — agree to disagree and move on.

It also tests the weights: how far a criterion's weight would have to shift before the winner changes. Small shifts flipping the result mean the weighting is worth a conversation; large ones mean it is robust.

Aim: reach a transparent, defensible decision quickly by spending discussion time only where it can change the answer.

Cost is deliberately left out of the scoring. Asking a group to score cost 1–10 is ambiguous — is 10 cheap or expensive? — and it buries real money inside a dimensionless score. So the MCA gives a pure benefit ranking, and the money comes back in here.

Enter capex and annual opex for each option and choose the basis: capex only, annual opex only, or whole-of-life present value (capex plus discounted opex). The tab then shows benefit per unit cost, flags any option that is dominated (costs more and delivers less — there is no argument for it), and works out the incremental cost of each extra benefit point along the efficient options.

The question this answers is the one that actually decides projects: the top-scoring option may only be marginally better — is that margin worth the extra spend?

Report builds a clean PDF or CSV of the sections you choose — laid out for printing with no browser header/footer and key blocks kept together across page breaks. Good for the decision record and audit trail.

Save writes a single JSON file with everything — decision, options, criteria, people and all scores — to your downloads. Load restores it. This is the only place to export, save or load, so keep that JSON as your backup.