The Gold Standard Reference | Diagnostic Review and Improvement Planning Workbook

Begin with a decision that has an owner and a time at which it matters. The learning-materials desk wants to know whether it can accept another weekly collection session. This focuses the review on dependable delivery and capacity, while still requiring attention to any other applicable core requirement that could invalidate the proposed claim.

Record the process, locations, people, period and method version within scope. Name what the review cannot establish. Observing one quiet session cannot establish performance during a busy handover; examining records alone cannot establish that a substitute can perform an unfamiliar task. Those limits should affect the conclusion and the next evidence request.

Distinguish a rapid diagnostic from a complete assessment. A rapid diagnostic may examine urgent uncertainties first. It must not present unexamined requirements as met. The useful output may be “defer expansion and examine these conditions”, even before every aspect of the system is fully assessed.

Workbook entry: State the decision, decision-maker, scope, relevant conditions, evidence deadline and limit of the intended conclusion. Explain what would require the work to pause immediately rather than wait for the diagnostic report.

Contents · Next chapter

Keep every core requirement visible. For each, link the full requirement to local criteria, relevant evidence and a finding. Subdivide an entry when its parts differ materially. A broad label such as “resources” should not conceal the difference between having enough kits and having a reliable booking record.

Start with “Unknown” where the evidence is insufficient. Unknown is a description of the assessment position, not a judgement that the organisation has failed. It creates a specific responsibility to examine the matter or limit the proposed decision. Marking an entry “Not applicable” requires a defensible scope-based reason; inconvenience is not a reason.

The example deliberately retains uncertainty. The missing acknowledgement may indicate a failed handoff or an incomplete copy of the evidence. Resolve that distinction before assigning a stronger conclusion.

Contents · Previous chapter · Next chapter

Choose cases for the questions they can answer. A routine completed case shows how the intended method operates. A delayed, disputed or interrupted case shows how the system handles difficulty. A case crossing a shift or location boundary exposes handoff assumptions. No fixed number of cases can guarantee useful coverage in every setting.

Follow each case from its initial promise to its final outcome and unresolved consequences. Compare the record, the operator's account and the relevant physical or system state. Note contradictions without immediately choosing the most convenient explanation. If records say a kit was ready but the participant received a substitute, ask which promise was actually accepted and when.

Sampling should have an explicit limitation. Deliberately choosing difficult cases is useful for finding weaknesses but does not estimate the overall frequency of those weaknesses. Conversely, a random-looking handful of records does not automatically justify a broad performance claim. See Evidence, Claims and Verification.

Workbook entry: List the cases, selection reason, complete evidence path and uncovered conditions. Record what further observation would change the interpretation. Stop collecting repetitive evidence when a different condition or unresolved contradiction is the more important gap.

Contents · Previous chapter · Next chapter

Different gaps require different responses. A criterion may be unclear, a control may be absent, a working control may be inconsistently used, or the control may operate without producing the intended result. A record may also be missing even though the work was completed. These situations should not all trigger the same training session.

At the learning-materials desk, one volunteer believes “collection complete” means the participant arrived; another believes it means the agreed materials were accepted. That is initially a definition problem. If a clear definition exists but an inaccessible screen prevents its use, access and process design need attention. If people understand and can use the rule but repeatedly bypass it under workload pressure, the review must examine the incentives and capacity as well as individual conduct.

Separate observations from hypotheses. “The shift was busy” is a condition. “High workload caused the missing handoff” is a hypothesis that needs comparison and examination. Several contributing conditions may operate together. Correcting one convenient explanation may leave the failure route open.

Workbook entry: Describe the gap in one sentence, identify what is known, list plausible explanations and choose the next discriminating check. Prefer the check that can distinguish between explanations over another general request for “more data”.

Contents · Previous chapter · Next chapter

Some findings require a protective action before a longer improvement plan is ready. If the desk cannot identify which kits are safe to release under its own criteria, it should isolate the affected work and establish a controlled decision route. A ranking exercise must not postpone that action.

After immediate protection, consider the consequence of failure, the exposure, the strength of current controls and the uncertainty. Then examine dependencies between improvements. Repairing the identity of each case may be a prerequisite for reliable handoffs and measurement. That dependency can make a modest-looking record change more important than a highly visible dashboard project.

These are planning categories, not new finding terms. The formal finding remains Met, Partly met, Not met, Unknown or Not applicable with justification. A priority label should never overwrite the evidence-based finding.

Contents · Previous chapter · Next chapter

“Train staff”, “update the process” and “communicate better” describe activities. An improvement plan also needs the mechanism expected to change the result. Write a short prediction: if this change is introduced under these conditions, this observable weakness should reduce, while these other outcomes remain acceptable.

For the desk, the proposed change is to require the incoming volunteer to reconcile open kit identities before accepting the queue. The prediction is that unowned discrepancies will be detected at the handoff instead of at the next collection. The team also watches handoff workload and delayed opening, so the new control does not quietly transfer the problem elsewhere.

Define what would count as disconfirming evidence. A completed reconciliation form is not sufficient if the next volunteer still cannot find the physical kit. The review should examine real transfers, including an exception and a substitute where those conditions matter. If the control works only with an experienced coordinator standing beside it, the release conditions must say so.

Workbook entry: Record the problem, proposed mechanism, owner, affected work, prerequisites, predicted result, balancing check, stop condition and review method. Set the review point from the opportunity to observe relevant work, not from an arbitrary promise that improvement takes a fixed number of days.

Contents · Previous chapter · Next chapter

A long action list can hide an impossible workload. Estimate the people, time, information and interruption each action needs. Include the work required to keep the current service dependable while the changes are tested. A plan that assumes improvement happens outside normal capacity often relies on invisible overtime or abandoned controls.

Build a dependency sequence. The desk first reconciles kit identities, then defines handoff acceptance, then tests the receiving check, and only then evaluates an additional session. These actions may overlap where evidence permits, but their dependencies remain visible. Training people in a method that is still changing may create avoidable rework.

Limit work in progress deliberately. The appropriate limit depends on the team's capacity and the nature of the changes; this programme does not prescribe a universal number. Record what is deferred, why, and which temporary controls remain necessary. Deferral without ownership can make an acknowledged gap disappear from attention.

Workbook entry: For each action, record prerequisite, committed capacity, accountable owner, next evidence event and consequence of delay. Review whether the same person has been allocated to incompatible commitments. Resolve the conflict before announcing a delivery date.

Contents · Previous chapter · Next chapter

Completion asks whether the promised action occurred. Effectiveness asks whether it produced the required change under the relevant conditions. Both matter, and they need different evidence. A revised instruction can be completed today while its effectiveness remains unknown until suitable work occurs.

At the desk, completion evidence includes the new agreement, briefing and available records. Effectiveness evidence includes observed handoffs, matched kit identities, receiving decisions and the treatment of exceptions. A review should also ask whether the change created delays or burdens that threaten sustainable use.

Do not silently change the success criterion after seeing the result. If the original test was unsuitable, explain why, approve a revised test and preserve the earlier result. A failed trial is useful evidence when it leads to a better design. Calling every trial a success deprives the team of that learning.

Workbook entry: Give each action separate completion and effectiveness states. State the examined conditions and unresolved limits. Close the action only when the required evidence supports closure, or explicitly replace it with a new decision that retains the unresolved issue. Failure, Correction and Recovery develops this distinction.

Contents · Previous chapter · Next chapter

A useful diagnostic report states what was examined, the important findings, the immediate controls, the proposed sequence and the decision now required. It should distinguish a finding about the current system from a prediction about an untested improvement.

For example: “The current desk can continue within the existing session limit while the discrepancy queue is reconciled. The evidence does not support an additional session. Handoff acceptance and substitute readiness remain unresolved. Reconsider expansion after the agreed trials and review of capacity.” This is a fictional reporting example, not a universal conclusion or a conformance statement.

Do not report a percentage of requirements met as if it settled the decision. A high percentage can coexist with a critical gap in ownership, evidence or readiness. Preserve every applicable requirement and explain how the unresolved findings affect the claim. “Unknown” should remain visible until evidence resolves it.

Workbook entry: Write the decision in plain language before attaching the register. Name who accepts the current limits, who must act and what evidence will reopen the decision. A reader should be able to identify the next responsible person without interpreting a colour code.

Contents · Previous chapter · Next chapter

Case: The fictional desk reviews twelve collections. Ten have complete issue records. Two contain only a volunteer's initials. All twelve participants eventually received materials. The manager proposes reporting a high completion rate and opening another session. The desk has not examined how late substitutions affected the participants or whether the substitute volunteer can reconcile returns.

Before reading on, write three separate conclusions: what the observations establish, what remains unknown and what the manager should decide next. Then choose one immediate action and one evidence request that would be most useful. Avoid turning twelve selected cases into an unsupported claim about all future sessions.

Reasoned response: The observations establish eventual receipt in the twelve reviewed cases and incomplete issue records in two. They do not establish that each agreed promise was fulfilled on the intended terms. The timing, substitutions and missing record meanings need examination. Substitute readiness also remains unknown. The evidence therefore does not yet justify the proposed expansion.

A proportionate next step is to reconstruct the two incomplete cases, check the consequences of substitutions and observe the substitute performing the complete return process. The team should examine capacity for the additional session alongside these findings. It should not infer that a better record alone proves additional capacity.

Use the register to connect each result to the applicable requirements, then issue a scoped decision. The workbook succeeds when the team can explain its sequence of actions and the evidence that would change its mind. Filling every cell is not the goal; a defensible improvement decision is.

Contents · Previous chapter · Continue through the reference program