Scoring Conduct, Not Just Completion: A Supplier Review Framework for UK Programmes

The limits of a rear-view scorecard

A supplier can meet every milestone on a UK programme and still leave the client with a fragile operating model. The scorecard says green; the delivery team knows otherwise. Traditional supplier scorecards tend to reward what is easy to count — on-time, on-budget, no open defects — while saying little about how the supplier behaves when the programme is under pressure. Those behaviours determine whether delivery holds together after the programme board has moved on.

The problem is not measurement itself. It is the choice of measures. On-time and on-budget are lagging indicators. They are also negotiable: a supplier that forecasts conservatively and then meets its own forecast scores well, while a supplier that flags risk early and absorbs disruption can score worse. The scorecard ends up rewarding the wrong conduct — caution in forecasting, silence on emerging issues, and a preference for completion over handover.

A behaviourally useful scorecard looks forward. It measures the signals that predict delivery outcomes rather than merely recording them. Four are worth adopting in most UK supplier relationships.

Decision latency

Decision latency is the elapsed time between a decision request being raised and a named owner making a dated decision. Measure the median and the 90th percentile, and track the ageing of open decisions. The median tells you how the relationship normally works; the 90th percentile tells you what happens when a decision is difficult.

Why does it matter? Slow decisions consume contingency and erode trust. A supplier that surfaces decisions early, with options and consequences, behaves differently from one that waits for the client to notice. The measure also exposes client-side bottlenecks. In many programmes, the supplier is blamed for delay when the real constraint is an unresolved client decision.

Evidence should come from a decision log, programme board minutes and requests for information. Avoid self-reported averages; sample the log and check that decisions are genuinely closed.

Forecast reliability

Forecast reliability is not the same as on-time delivery. It is the stability and accuracy of the supplier's forecasts over time. Track how often milestone dates change, the size of each change, whether the supplier is systematically optimistic or pessimistic, and how much notice is given before a change.

A reliable forecast is a management tool. An unstable one is an early warning. Score accuracy at 30, 60 and 90 days out, not only at final delivery. A supplier that gives early notice of a slip is more useful than one that protects the date until the last possible moment. The scorecard should reflect that difference.

Rework rates

Rework is where programme cost and schedule quietly disappear. Measure the proportion of deliverables rejected at first review, reopened after acceptance, or subject to concession. Categorise root causes: requirement ambiguity, design error, interface mismatch, quality control. Record the cost of rework where it can be established.

First-time-right performance is a better indicator of delivery discipline than a completion date. A supplier that completes on time by relying on concession and rework has not delivered cleanly. The scorecard should make that visible.

Knowledge transfer

Knowledge transfer is often asserted in proposals and rarely measured in delivery. Build the measure around evidence: handover artefacts accepted, shadowing completed, competence sign-off by client staff, query volume after handover, and residual dependency on supplier personnel.

A handover ledger agreed at contract award is useful here. It converts knowledge transfer from a closing statement into a managed deliverable. Score against the ledger, not against a narrative in the final report.

Building a scorecard that is used

Use a simple 1–5 rubric for each signal, with written anchors. For decision latency, for example, a 5 might mean a median under five working days and a 90th percentile under fifteen; a 3 might mean a median of ten days; a 1 might mean a median over twenty days or no reliable log. The anchors should be agreed with the supplier before the first review.

Keep the number of measures small. Four behavioural signals plus the conventional cost and schedule measures are enough for most quarterly reviews. Adding more usually adds data burden rather than insight. Publish definitions at contract award, baseline in the first quarter, and judge trend rather than absolute score.

Do not over-weight the composite. A single percentage can hide the one signal that matters. Use the score to structure the conversation, not to replace it.

The quarterly review

The review should have a pre-read: the supplier's self-assessment with evidence, followed by the buyer's assessment. The meeting itself should focus on two signals, not all four. Ask what changed, what will change, and who owns the action. Record actions with dates and escalate unresolved items through the commercial route.

The scorecard only changes behaviour if it has consequences. That does not mean payment deductions in every case. It means the score informs decisions about future work, contract mechanisms and senior engagement. A score that is filed and forgotten is another reporting ritual.

Avoiding the ritual

Four failure modes are common. The first is data burden: too many measures, collected manually, reviewed late. The second is gaming: definitions that allow a supplier to score well without changing conduct. The third is blame: a review that punishes transparency and encourages silence. The fourth is false precision: a decimal score that implies certainty the evidence does not support.

Mitigate these with sampled audits, joint data validation, published definitions and a forward-looking agenda. Keep the scorecard proportionate. Its purpose is clearer delivery, not a fuller spreadsheet.

A modest test

Before rolling out a behavioural scorecard across a supplier portfolio, test it with one supplier and one programme for two quarters. Ask three questions: did the score change a decision, did it surface a risk earlier, and did the supplier's conduct change? If the answer to all three is no, simplify the scorecard or stop using it.

The measure of a good supplier scorecard is not how it looks in the board pack. It is whether it changes what happens on the programme. Clearer delivery begins with measuring the things that shape it.