Your primary system is unavailable for six hours. What must the business be able to do without it, and what have you prepared?

What is being tested

Whether you think beyond technical recovery to how the business keeps operating — which is the larger and more often neglected question.

What must be established first

A business impact analysis. For each business function, what is the impact of losing it for one hour, one day, one week? That produces the maximum tolerable outage, from which technical RTO and RPO are derived rather than invented.

Impact is not only revenue: regulatory obligations, safety, contractual penalties and customer trust all count, and some dominate.

The output is a prioritised list, which is the whole point. Not everything can be restored first, and deciding the order under pressure is not possible.

What must be prepared

1. Manual workarounds. How does the business operate for six hours without the system? Frequently the highest-value and cheapest continuity control, and frequently undocumented because it is unglamorous.

Retail can process sales on paper and reconcile later. A contact centre can record cases manually. Warehouses can pick from printed lists. These procedures must exist, be current, and — critically — have been practised, because staff who have never used them will improvise badly.

2. A reconciliation path. Every manual workaround creates a backlog to enter afterwards. That process must exist too, and it must handle the conflicts that arise when the system's state and the manual record disagree.

3. Communication. Who tells customers, staff and regulators, through which channel — noting that the usual channel may itself be affected. A status page hosted in the same infrastructure is a recurring and predictable failure.

4. Decision authority. Who decides to invoke continuity procedures, and who can be reached at 3am. Without a named person, the decision waits.

5. The plan itself, accessible offline. A continuity plan stored in the system it covers is the classic failure.

The dependencies people miss

Third parties. A payment provider, a delivery partner, a key SaaS application. Your plan is incomplete if it assumes suppliers are available, and their availability is not something you control.

The control plane. Identity, secrets, deployment and observability. The recurring finding across the industry is that these fail the test — organisations discover during a rehearsal that their planned response is impossible because the tooling to execute it shares the failure.

What makes it real

Rehearsal, including the non-technical parts, with the participants who would actually be involved: support, communications, and whoever approves customer statements. No document review produces the control-plane finding; only attempting it does.

What a strong answer adds

That the honest answer to "what must the business be able to do" is usually a short list — take orders, serve existing customers, meet a regulatory obligation — and that focusing continuity investment on that short list is far more effective than attempting to preserve everything.