Governance

A decision record that holds up to scrutiny

When a manager accepts a risk, the next team needs to know why. Record what was agreed, who must act and what would require a new decision.

When you approve delaying a security repair, the people running the affected service need to know more than the date of the meeting. What can they continue doing? Which restrictions must stay in place? Who will pay for and complete the repair? Without that information, a new manager may inherit a temporary arrangement and assume it was intended to be permanent.

Write down the decision so someone who was not in the room can act on it. That record should explain what could still go wrong, why the chosen action was reasonable and what would require a fresh decision. It must also distinguish choices the business can make from legal requirements, promises to customers and insurance terms. Showing that a repair costs more than an estimated loss does not establish that postponing it is lawful or permitted by a contract.

Check what the organisation is required to do

Before comparing options, establish which company in the group provides the service, where it operates and which contracts cover it. Ask qualified legal advisers or the person responsible for compliance to determine the requirements. The EU cybersecurity directive commonly called NIS2 is one example where assumptions about coverage can mislead. The European Commission’s NIS2 guidance describes sector and size criteria, alongside exceptions. A company’s industry label alone does not establish whether it is covered. Its circumstances and the relevant national legislation need checking.

For each requirement, record where it comes from, who interpreted it and what it means for the proposed decision. Make clear whether it is a legal duty, a rule the organisation has set for itself or a commitment to a particular customer. If the answer is not yet known, give someone responsibility for resolving it before an option that depends on that answer is approved. The manager deciding the budget should not discover afterwards that the cheapest approach was never available.

Record what people must do after approval

Begin with the work affected and the choice being made. Explain what could fail and what each available option would mean for staff and customers. Keep the record short enough to use, but include enough information for a later reader to understand the decision using the evidence available at the time. State what was not known.

  1. What is affected: the service, information and customers involved, plus the systems or suppliers needed to keep it working.
  2. What we know: the dated findings, what was examined, assumptions and unanswered questions.
  3. The choices: what can be done, what it costs, any restrictions on how work can continue and what could still go wrong afterwards.
  4. Who can decide: the person accountable for approval and the engineering, operations or legal advice needed before they give it.
  5. What happens next: the funded work, who will deliver it, when it is due and what would make the decision need reviewing.

Link to reports held where only authorised people can read them, rather than copying sensitive details into every meeting paper. Identify the report version, date and what it actually examined. If engineers tested one way of accessing a system, say so. Repeated summaries must not turn that limited test into a claim that the whole service is secure.

Explain what a red rating means

A colour or score can help a committee find decisions needing attention, but it cannot replace the explanation. A “high risk” rating might mean that work could stop for a long time, that confidential information could be stolen or that nobody has checked whether a protective measure works. These call for different actions: testing recovery, restricting who can see the information or investigating the protection. Give the rating a definition and a link to the actual problem.

Apply the same care to financial estimates. State the period covered, the possible range of loss and significant costs left out. Distinguish measured facts from judgement. If changing an assumption would reverse the choice, show that plainly. A calculation of possible outcomes is no more reliable simply because its results have several decimal places.

Put conditions and an end date on temporary permission

Consider a hypothetical approval to keep an older service running while an upgrade is prepared. The approval depends on only named partners being able to use it. The record needs to say who keeps that restriction working and how the team checked it. Adding another partner or allowing a different connection should prompt review, because the original decision did not cover that change.

Specify exactly what the permission covers, when it expires and who is responsible for it. Say who must be contacted if work is late or the restrictions cannot be maintained. Extending the deadline requires current evidence and a decision about the unfinished work, not an automatic date change. Keep earlier versions so a new manager can see whether a delay was deliberately approved or a promised repair simply did not happen.

Use the record when something goes wrong

After an incident, compare what happened with the failure considered when the decision was made. Was an assumption wrong? Was a promised protective measure never installed, or did it stop working? Did a change that should have prompted review go unreported or unaddressed? The answers matter: better analysis will not fix a failure to deliver approved work, and completing that work will not correct an unrealistic assumption about how the business operates.

At the next review, use the record to check what has changed, which repairs remain unfinished and whether the temporary restrictions still work. Confirm who is taking the next action. An approval filed away and never revisited can leave staff running a service on terms that no longer match reality.

Where this applies in our work: the quantification method.