Weighting and scoring is the method the guidance warns against

Most option appraisals arrive at the same artefact. A table. Options down the left, criteria across the top, a weight against each criterion, a score in each cell, a weighted total, and a row that wins.

It looks like analysis. It produces a single defensible-looking number per option, it can be presented in one slide, and it converts a hard argument into arithmetic.

The Green Book identifies the technique by name and advises against it.

The Green Book recommends against the use of multi-criteria analysis (MCA) in decision making. MCA, which is distinct from MCDA, involves simple weighting and scoring that lacks an objective basis and can reduce transparency.

Two objections, and they are worth separating.

No objective basis means the weights and the scores came from somewhere unrecorded. Someone decided that deliverability is worth twenty per cent and whole life cost fifteen, and that this option scores four on the first. None of those numbers is derived from evidence, and none is written down as a judgement either, because the format has no place to put reasoning.

Reduces transparency is the sharper one. The table does not hide the result; it hides the argument. You cannot disagree with a weighted total. You can only disagree with a weight, and by the time the total exists nobody remembers where the weight came from. A number that cannot be argued with specifically has removed the thing appraisal was for.

Note the distinction the guidance draws. Multi-criteria decision analysis, a different method, is endorsed for complex technical trade-offs, using swing weighting to weigh informed expert and stakeholder opinion, and with an expert facilitator. So the objection is not to structure. It is to weights that nobody has to defend.

What the guidance recommends instead

The recommended method is the options framework filter, and its underlying idea is not a better ranking. It is a different account of where options come from.

The usual assumption is that options exist first and appraisal chooses between them. In practice this means somebody arrives with three or four solutions already in mind, and the appraisal is the ceremony that selects among them. Anything nobody thought of before the meeting is not in the table, and the table cannot show its own omissions.

The filter inverts this. It decomposes the proposal into categories, identifies the possibilities within each, discards the ones that fail, and then generates options by recombining what survives.

The vocabulary matters because the three words get used interchangeably. An option is a potential way of meeting the objectives. A category is an attribute that varies between options, such as building material. An option choice is one possibility within a category, such as timber frames rather than steel.

The Green Book recommends five categories, taken in order:

• Scope. What is the coverage of the service to be delivered? Varied by geography, population, quality or duration.
• Solution. How can the outcomes be delivered? New infrastructure, refurbishment, regulation, taxes, grants, information campaigns.
• Delivery. Which organisation will deliver it? Direct provision, partnership, not-for-profit, private sector.
• Implementation. How will it be rolled out? A pilot, a phased rollout, rapid implementation.
• Funding. How will it be paid for, given the preferences settled in the four categories above.

Each category is worked from the minimum that meets the objectives to a much more ambitious maximum, with the possibilities in between considered rather than assumed away. Each is rated against the critical success factors: green for the choice that meets them best, amber for one that meets them less well, red for one that does not and is discarded. Options are then generated by combining the green and amber choices.

The sequence is deliberate. Solution is considered against the green scope, delivery against the green scope and solution, and so on, so that each decision is taken with the previous ones fixed rather than everything floating at once.

The two features the guidance calls essential

Both are about process rather than analysis, and both are the first things dropped.

The first is structured facilitated workshops, where the professions and stakeholders go through the method together rather than receiving its output. The second is explicit deliberation: participants must explain the evidence, research and assumptions behind their views, and the reason for every rating must be recorded in a short paragraph naming its underlying evidence and sources.

That recording requirement is what the weighted scoring table cannot do. It is also, not coincidentally, the thing that makes the appraisal auditable a year later, when somebody asks why an obvious option was never considered.

The same requirement, arriving from the asset management side

, read: ISO 55001 now requires you to write down how you decideISO 55001 now requires you to write down how you decide

Critical success factors, which do the filtering

The filter needs something to filter against, and that is the critical success factors: the attributes an option must have to achieve the objectives successfully. The guidance sets out five that apply to every proposal.

Objectives and strategic fit, meaning the option meets its SMART objectives and fits what else the organisation is doing. Value for money. Supplier capacity and capability, including whether a mutually acceptable agreement on payment and risk transfer is reachable. Affordability, for all of the public bodies involved. Achievability, meaning arrangements exist to deliver, monitor and evaluate it.

Constraints belong here too: legal requirements, ethical standards, social acceptability, timing, emissions policy. A constraint expressed as a critical success factor removes options early and cheaply. The same constraint discovered during shortlist appraisal removes months of work with them.

Five options, and why five

The shortlist should generally contain around five. The guidance gives its reason plainly: a range of that size means decision makers are presented with genuine choices rather than a single pre-determined outcome.

The standard five are business as usual, do minimum, a preferred way forward, a more ambitious version of it and a less ambitious version. The last may be dropped where the preferred way forward is itself the do minimum.

Two of those five are controls rather than candidates. Business as usual is the outcome if nothing is done. It need not meet the objectives and is carried to shortlist appraisal regardless, as the benchmark for whether to act at all. Do minimum does have to meet them, at the smallest scope that will, and exists to test whether the more ambitious options earn their extra cost. The guidance has a word for what the comparison catches: gold plating.

An appraisal that carries neither has no test of whether to act and no test of how much.

The stage most often skipped

The guidance is unusually direct about this, and the sentence is worth reading as a description of normal practice rather than a warning.

Options generation and longlist appraisal, it says, are important but are often overlooked. Practitioners are often tempted to skip these stages and progress straight to shortlist appraisal and the detailed analysis of costs, benefits and risks.

That temptation is easy to understand. Shortlist appraisal produces numbers, and numbers look like work. Longlist appraisal produces a documented account of what was considered and rejected, which looks like overhead right up until somebody asks the question it answers.

What the Treasury names as the largest single cause of overrun

, read: The largest contributor to optimism bias is the document that starts the projectThe largest contributor to optimism bias is the document that starts the project

---

Sources. HM Treasury, The Green Book, 2026 edition. Paragraph 5.15 for the position on multi-criteria analysis and 5.14 for multi-criteria decision analysis and swing weighting. Paragraphs 5.21 to 5.32 for the options framework filter, its categories, its ratings and the sequence in which categories are taken. Paragraphs 5.22 and 5.23 for facilitated workshops and explicit deliberation, and 5.27 for the requirement to record the reason for each rating. Paragraph 5.5 and Table 4 for the critical success factors, and 5.6 for constraints. Paragraphs 5.17 and 5.18 for shortlist size and the five options. Paragraph 5.3 for the observation that options generation and longlist appraisal are often skipped.