Non-price criteria need disclosed meaning

A cost comparison cannot determine how much an organisation values a service difference that its cash model does not represent. Quality, delivery, support, adaptability and other attributes may be relevant to an award. They need disclosed meaning and an authorised role. A label such as quality is too broad to evaluate consistently: it might refer to technical compliance, documented capability, useful service beyond the minimum or confidence in a promised outcome. Those are different propositions with different evidence and consequences.

Begin by deciding which conditions are mandatory. A requirement that determines admissibility should be assessed before compensating preferences. A supplier that cannot meet an overriding safety or essential-service condition should not regain eligibility merely by offering a lower cost or an attractive unrelated feature. The evaluator should retain the requirement, evidence and authorised treatment of an unresolved condition. A cost engine can calculate an inadmissible alternative's figures, but that ability does not give the alternative a place in the final award comparison.

Separate compliance from a valued increment. Meeting a required capacity or documentation condition can establish the minimum perimeter without creating an additional quality advantage. A feature beyond that perimeter may be valuable if the buyer has a defined use for it. The evaluator should identify that use and the relevant evidence. Rewarding an impressive specification without a service consequence can purchase unused capability. Equally, treating every above-minimum capability as neutral can overlook adaptability or continuity that the organisation has explicitly authorised itself to value.

An evaluated attribute should have a direction and a reference basis. A delivery assessment can distinguish a committed usable-service date from an indicative dispatch date. A support assessment can distinguish demonstrated restoration capability from a list of offices. An interoperability assessment can distinguish an implemented compatible interface from a general promise of integration. Without those definitions, evaluators can assign different meanings to the same score while appearing to use a common matrix. The resulting aggregate then obscures disagreement rather than resolving it.

Evidence should match the criterion. A relevant operating record can support a service-capability proposition while requiring adjustment for the proposed duty. A technically sound plan can support understanding of the requirement without proving future delivery. A certification can have a scope narrower than the offered integrated service. The evaluator should state what each item establishes and avoid transferring confidence across unrelated propositions. Non-price evaluation needs the same provenance discipline as an annual cost input, even when the final output is qualitative.

Choose an evaluation architecture before inspecting which option it favours. One approach can retain mandatory gates and then compare cost among qualifying offers. Another can compare disclosed preferences alongside cost under an authorised decision rule. A committee can also retain separate dimensions and deliberate explicitly on the trade-off. These architectures answer different governance questions. The analyst should not switch between them after seeing the results, because doing so can change what the invitation appeared to reward and make the evaluation difficult to challenge.

Cost and non-price attributes should be reconciled for overlapping mechanisms. If lower electricity has already reduced lifecycle cost, rewarding that same cash saving again as a broad efficiency score can double count its economic effect. There may still be a separate authorised objective, such as a non-cash environmental consequence, but it needs a distinct definition and evidence. If faster restoration has already reduced the downtime allowance, an additional support preference should identify what service benefit remains outside that allowance. The question is whether the two criteria evaluate different consequences.

A qualitative preference need not be forced into money to be legitimate. Some decision owners may lack authority or evidence to assign a monetary value to adaptability, delivery confidence or a particular operating condition. They can retain the attribute with an explicit meaning and decision role. The limitation is that the cost engine cannot aggregate it. A report should show the cash-cost result and the separate evaluation rather than claim the model has computed best overall value. A disclosed judgement can be more defensible than an invented monetary penalty.

Weighted scores require a coherent scaling and trade-off interpretation. A numerical scale can make preferences look objective while depending heavily on definitions, anchors and weights. An improvement between adjacent score labels may not have the same service meaning across criteria. A large weight can also compensate for a condition the organisation intended to treat as mandatory. If a separate scoring method is used, it should be verified and its sensitivity retained. The procurement workbench in this paper contains no weighted-quality calculation or automatic cost-and-quality award engine.

Test the proposed criteria with contrasting hypothetical offers before using them for a consequential evaluation. The test should ask whether a plainly inadequate mandatory condition can survive compensation, whether the same benefit is counted twice and whether a superficially impressive but unusable feature is rewarded. It should also ask whether evaluators can distinguish a documented capability from an unsupported promise. Such tests can expose a flaw in the architecture before the buyer has actual bids and a temptation to interpret the rules around a preferred outcome.

Evaluator disagreement is useful evidence about unclear meaning. If one evaluator reads support as local presence and another reads it as demonstrated recovery, averaging the scores does not resolve the issue. The evaluation lead should identify the definition, reconsider the supporting evidence and apply the same clarified meaning through the authorised process. A disagreement can also reflect a real trade-off between service objectives. That trade-off should be recorded for the decision authority, rather than compressed into a numerical average that conceals its governance significance.

Use a reasoned record with the score or qualitative conclusion. It should identify the evidence, the criterion anchor and the assessed consequence. A strong score with no explanation is difficult to reproduce; a long narrative that does not apply the anchor can be equally weak. The record should let another evaluator see why the evidence meets the particular level or condition. It should also retain unresolved qualifications. An indicative proposal can remain relevant without being described as a committed capability.

Beware of criteria that reward presentation rather than execution. A polished response may explain the proposed method clearly, which is useful, but its appearance does not establish capability. A long list of technologies can similarly create an impression of breadth while leaving the required interface unimplemented. The evaluator should follow the chain from proposal to relevant evidence and service consequence. Where the criterion intentionally values understanding or planning quality, its meaning should remain separate from demonstrated delivery.

Sustainability or resource criteria need a defined boundary. An organisational claim can differ from the performance of the offered service. A component's attribute can differ from the lifecycle effect of the whole delivery model. A purchaser may legitimately use a distinct authorised non-cash objective, but it should identify the proposition and evidence rather than attach a broad favourable label. The cost model includes only its defined energy cash effect; it supplies no emissions inventory, social-impact measurement or automatic sustainability score.

Use a dominance challenge before applying weights. If one hypothetical offer is no worse on every disclosed attribute and better on a material one, a score that prefers the other deserves investigation of scale, direction or duplicated criteria. If the offers trade advantages, the authority must own the trade-off rather than hide it inside unexplained points. A service attribute already monetised in interruption cost should not receive an additional benefit merely because it also has a reliability label. A mandatory condition should not be diluted by attractive scores elsewhere. This challenge tests whether the evaluation architecture behaves as declared while leaving the institution's legitimate preference judgement visible.

Evaluator disagreement can be resolved by separating observation from preference. In a hypothetical support criterion, evaluators may agree that local capability is absent but disagree about its importance. More evidence on office locations cannot resolve that value judgement. They may instead disagree about whether the proposed subcontractor is available for the duty; that is an evidential dispute requiring a capability check. Record the shared observation, disputed proposition and authorised preference separately. A revised factual conclusion can change the evaluation without changing weights; a revised mandate can change weights without altering the observed service. Keeping those branches distinct makes the resulting preference explainable without relabelling the lifecycle cost leader as an automatic overall winner.