Weighted criteria scoring: how to compare project options fairly

Most projects have more than one sensible way forward. Cost is rarely the only thing that matters, so teams need a fair way to weigh cost against risk, deliverability, safety and everything else they care about. Weighted criteria scoring, also called multi-criteria analysis or a weighted scoring matrix, is the most common way to do it.

The idea is simple. Agree what matters, agree how much each thing matters, then score every option against the same criteria. Done well, it turns a debate about opinions into a discussion about evidence. Done badly, it produces a precise looking number that nobody trusts.

1. Start from the needs case

Criteria should come from what the project is trying to achieve, not from the options already on the table. If the needs case says the site must stay compliant, keep running during the works and last another twenty years, those are your first clues. Write down the objectives before anyone mentions a preferred solution. See our guide on writing a clear needs case.

2. Separate the gates from the scores

Some requirements are not trade offs. An option that breaks the law, fails a safety standard or cannot be funded should not be rescued by scoring well elsewhere. Treat these as pass or fail gates (the HM Treasury Green Book calls them critical success factors) and apply them before scoring. Only options that pass go forward to the weighted comparison.

3. Choose three to eight criteria

More is not always better. Three well chosen criteria applied consistently, for example cost, resultant site risk and regulatory benefit, often work better than a long list that makes the winner harder to see. More than eight and each one carries so little weight that the scoring becomes noise. Good criteria are distinct from each other, relevant to the decision and capable of being evidenced. Common ones for engineering projects include:

• Capital cost: the upfront cost to deliver.
• Cost performance: cost against budget or target.
• Whole life cost: capital plus operating, maintenance and replacement costs over a set period.
• Resultant risk: the risk left once the option is delivered.
• Strategic alignment: fit with long term plans, standards and policy.
• Regulatory benefit: contribution to regulatory outputs and commitments.
• Deliverability: programme, outages, access and dependencies.
• Safety and environment: safety, carbon and environmental impact.
• Operability and maintenance: how easy it is to run, inspect and maintain.
• Stakeholder impact: disruption, service levels and public perception.

Watch for double counting. The three cost measures overlap, so usually choose one, or pair capital cost with cost performance. Otherwise the same money counts more than once.

4. Set the weights before anyone scores

Weights express how much each criterion matters, usually as percentages that add up to 100. Agree them with the key stakeholders before the options are scored. Weights chosen after the scores are known tend, consciously or not, to favour someone's preferred option.

Two simple ways to set them:

• Points allocation. Give each stakeholder 100 points to spread across the criteria, then discuss the differences and agree a set.
• Pairwise comparison. Compare criteria two at a time ("is cost more important than programme?") and let the pattern of answers set the weights.

Record the reason for each weight. A reviewer will ask why cost is 30% and not 50%.

5. Define what each score means

A score of 4 means nothing unless everyone agrees what a 4 looks like. Write a short descriptor for each point on the scale, per criterion.

Some teams prefer red, amber and green ratings. They are quicker and easier to present, but they flatten differences. A common compromise is to score on a 1 to 5 scale and show the result as colours.

6. Score with evidence, then compare

Score each option against each criterion and write down the evidence behind every score: the estimate, the survey finding, the programme, the specialist's view. Where possible, ask people to score on their own first and then discuss. It surfaces genuine disagreement instead of letting the loudest voice settle it.

The weighted score for each option is the sum of each score multiplied by its weight. Here is a worked example with scores out of 5, converted to a total out of 100:

Option B comes out ahead. Upgrading the existing unit is cheapest to build, but the ageing equipment costs more to run and maintain, so here the new unit has the lower whole life cost. That will not always be the case, which is why every score needs its evidence. Option C scores best on strategic alignment because it adds future capacity, but it costs more and takes longer.

7. Test the result

Before presenting a winner, check how sensitive it is. Change the weights within a reasonable range and see whether the ranking holds. In the example above, if the organisation placed more value on future capacity, moving 15 points of weight from whole life cost to strategic alignment, Options B and C would draw level on 82. The decision then rests on how much the organisation values future growth against cost. That is the conversation the decision makers need to have.

8. Record the rationale

Alongside the numbers, record the benefits and trade offs of each option, each stakeholder's view and the reason the preferred option was chosen. If the scores and the decision ever disagree, the recorded rationale is what explains it. Months later, it is often the only part anyone reads.

Common mistakes

• Weights set after scoring, or adjusted until the favoured option wins.
• Too many criteria, so each carries a trivial weight.
• Double counting cost or risk under two different names.
• Scores without evidence, which cannot be defended when challenged.
• Pass or fail requirements scored instead of applied as gates.

How DevelopVault helps

DevelopVault keeps the criteria, weights, scores and the evidence behind each score in one visual record, alongside the options themselves. Benefits, trade offs and stakeholder votes sit next to the numbers, so the decision and the reasoning stay together. See DevelopVault Pro.

Frequently asked questions

Related guides

This guide is general information, not professional or regulatory advice.