Nuage Solutions All articles
Cloud Strategy

Cover Before the Vacancy: A Key-Person Audit for UK Cloud Estates

Nuage Solutions
Cover Before the Vacancy: A Key-Person Audit for UK Cloud Estates

Most cloud reviews begin with architecture diagrams, contract terms and security settings. Few begin with a simpler question: who, exactly, can operate this estate if one person is unavailable? That omission is understandable. Cloud platforms are designed to abstract infrastructure, so it is easy to assume they also abstract people dependency. They do not. Decision rights, break-glass access, tenant administration and vendor relationships still sit with named individuals, and when those individuals leave, the estate does not fail immediately. It loses the ability to respond.

For UK organisations planning cloud operations through 2026, staff turnover should be treated as a continuity risk in its own right. The issue is not disloyalty or poor documentation alone. It is the gap between what an organisation believes it can recover and what it can actually operate when the usual person is on leave, unwell or serving notice. A lightweight key-person audit, followed by a tested handover routine, closes much of that gap without turning continuity into a paperwork exercise.

Why key-person risk hides in the cloud estate

Cloud platforms concentrate authority in portals and role assignments. A small number of administrators may hold privileged roles across Microsoft 365, Azure or a similar environment. A finance contact may approve tenant-level changes or subscription amendments. A contractor may have built the automation that keeps a backup schedule running, but the only explanation of how it works may live in their own notes. None of this appears on a typical organisation chart.

The risk is often described as a documentation problem. In practice, it is a cover problem. A runbook can exist and still be unusable if no one else has walked through it under realistic conditions. A vendor contact can be listed and still be unknown to the person who would need to call them during an incident. The test is not whether knowledge has been written down, but whether a second person can act on it without the first person present.

Four knowledge classes worth mapping

A useful audit avoids trying to capture everything. It concentrates on the knowledge that would slow down recovery, or stop it altogether, if a key person were unavailable.

Access and authority. This covers break-glass accounts, privileged administrator roles, tenant-level settings, billing and subscription approvals, domain registrar access and support contract ownership. For each item, record who holds it, who else could obtain it, and what evidence exists that the second person has used it.

Process knowledge. This includes restore and failover procedures, mailbox recovery, leaver processing, licence changes, incident escalation and change approval. Processes that are performed rarely are often the most fragile, because the primary operator may be the only person who has seen them work.

Relationship knowledge. Vendor account managers, finance approvers, legal contacts, the data protection lead and cyber insurance contacts all matter during an incident. A name in a contract is not the same as a working relationship. If the secondary contact has never spoken to the vendor, the cover is theoretical.

Context knowledge. This is the most easily missed category: why a configuration exists, which exceptions are deliberate, where a workaround is still in use, and which dependencies are undocumented. Context is what prevents a new administrator from making a technically correct change that breaks a business process.

A half-day audit that produces evidence

A key-person audit does not need to become a large programme. Start with the services that would cause the greatest operational or regulatory impact if they could not be administered for a week. For a UK organisation, that may include identity services, email, finance systems, backup platforms and any service that supports statutory reporting or customer obligations.

For each service, ask three questions. Who can perform the critical tasks today? Who else could perform them if the first person were unavailable? What evidence shows that the second person has done so successfully? Record the answers in a simple register alongside the last test date. The register should include contractors and third-party managed service contacts, not only permanent employees.

Prioritise by impact and time to restore, not by job title. A senior manager may sponsor a service without holding any operational knowledge, while a relatively junior administrator may be the only person who knows how to restore a deleted shared mailbox. The audit should surface those asymmetries before a vacancy does.

Test cover without disrupting the business

Cover arrangements should be exercised in controlled conditions. A practical method is reverse shadowing: the secondary person performs a routine but important change while the primary person observes. Another is a short game day in which the secondary administrator follows the restore runbook for a non-production workload and records where the instructions fail.

Break-glass access deserves particular attention. It is often held by one or two people and used rarely. A controlled rehearsal, scheduled in a change window with audit logging enabled, can confirm that the process works and that the second person can follow it. The aim is not to create drama; it is to replace assumption with evidence.

If someone has already given notice, test cover during the notice period rather than after departure. Have the secondary person complete a live task with the leaver available only for questions, not for hands-on work. This exposes gaps while they can still be corrected.

A handover routine that survives the notice period

Handover should begin when notice is received, not in the final week. UK notice periods vary, but even a month is short if access, context and relationships all need to move. A named recipient should own the handover, with a checklist covering access transfer, runbook walkthroughs, vendor introductions, outstanding changes, known issues, renewal dates and contract obligations.

Each item should have evidence attached: a completed change, a test result, a meeting note or a signed acknowledgement. Access should be transferred deliberately and revoked promptly once the handover is complete. An exit conversation can also capture context that formal documentation misses, particularly where a long-standing workaround has become part of normal operations.

Make the review routine

Key-person risk changes with every joiner, mover and leaver. A quarterly review is usually enough for a small UK estate, provided it is linked to HR and managed service review cycles. A useful board-level measure is the proportion of critical services with a named secondary operator and a test completed in the last twelve months. That single metric says more about operational resilience than a shelf of untouched documentation.

The objective is not a binder that proves knowledge was captured. It is confidence that someone else can act. A cloud estate is only as resilient as the second person who can operate it, and that second person should be identified, briefed and tested before the vacancy appears.

All Articles

Related Articles

The Residual Estate: A Half-Day Audit of What Cloud Migration Left On-Site

The Residual Estate: A Half-Day Audit of What Cloud Migration Left On-Site

The Exit File: How UK Organisations Can Prove a Cloud Service Has Been Retired

The Exit File: How UK Organisations Can Prove a Cloud Service Has Been Retired

Consent That Outlives the Project: A Quarterly Audit of Third-Party App Access

Consent That Outlives the Project: A Quarterly Audit of Third-Party App Access