An inventory of laptops and applications is useful, but it doesn't explain the whole of what a business needs to protect. I would begin with a different question: what must continue to happen for customers to be served, commitments to be met and the business to be paid?

Looking at business services rather than individual systems brings less obvious dependencies into view. An application might be running but unavailable to staff because the sign-in service has failed. A platform might be restored while its records are too incomplete to trust. Neither outcome is particularly helpful to the people trying to keep the business operating.

Describe a service in business terms

A service is an outcome the organisation delivers: accepting orders, arranging engineers, paying staff or producing management accounts. It needs an owner who understands the consequences of disruption and can decide whether a recovered service is fit for use.

There is no need to map the entire organisation before taking action. I would start with the small number of services whose interruption would have the most immediate impact on operations, customers or finances. A mapping exercise that becomes too elaborate can consume effort before a single important assumption has been tested.

The NCSC's incident preparation guidance recommends identifying essential information and critical processes as a basis for response planning [S2]. A service map connects those requirements to the people responsible for the business outcome.

I would ask the service owner to describe the outcome without naming the supporting technology. What initiates the work? What must be delivered? Who notices when delivery stops? Which obligations are time-sensitive? Once those questions are clear, the supporting applications, people and suppliers can be mapped in context. This matters because several systems may be technically healthy while an important business service is failing. Equally, a system outage may be manageable for a period if a genuine and sufficiently capable alternative exists. Not every technical failure has the same business consequence.

Follow the dependencies

For each service, work through what it actually needs: applications, data, user identities, network connections, suppliers and key people. Include the administrative and recovery arrangements too. For example, access to a backup platform may rely on the same identity service as the systems the business hopes to restore.

Consider an illustrative distribution company. Fulfilling an order could depend on a customer portal, current stock records, payment processing and a connection to the warehouse. Employees also need reliable delivery details. A manual alternative may be possible, provided those details remain accessible and someone can reconcile the resulting transactions later.

Simply recording “manual process available” would give false confidence. What does that process involve? How many orders can it manage? What information does it require, and for how long can it realistically continue? Until somebody has exercised it, the fallback remains an assumption rather than demonstrated capability.

Recovery dependencies deserve particular attention because some are invisible in normal operation. A privileged administrator, a recovery key, a current configuration record or an out-of-band supplier contact may not be used day to day, yet can become essential during restoration. A manual fallback can also quietly rely on the system it is supposed to replace. If warehouse staff need stock or delivery records held only in the unavailable platform, 'paper process available' is not a credible recovery statement. I would test the availability of the information and the capacity of the alternative, not merely confirm that a written procedure exists.

Examine availability, integrity and confidentiality

Three different problems are easy to confuse. Availability is about having access to a service or information when it is needed. Integrity is about its accuracy and trustworthiness. Confidentiality concerns who is permitted to see it. A plan focused solely on keeping systems online will miss some important consequences.

An order platform could remain available while delivery addresses are altered without permission. The finance team might be able to open supplier records but no longer trust the bank details. Confidential files could be copied without interrupting normal operations at all.

The service map should therefore describe which consequence matters in each case. Recovery might require reconciling transactions rather than merely restoring an application. Protection might mean improving an approval process or removing unnecessary access, not buying another security product.

These distinctions lead to different kinds of assurance. An availability test asks whether the necessary service can return within an acceptable period. An integrity check asks whether altered or inconsistent data can be found and corrected. A confidentiality review asks who can access sensitive information and how inappropriate disclosure would be handled. Passing one test does not demonstrate the others. For smaller businesses, I would keep the description concise but specific: what information must be present, how will employees know it can be trusted, and who should be able to see or change it?

Separate the requirement from the tested result

Two measures help frame a recovery discussion. The recovery time objective (RTO) is the intended time to restore a service. The recovery point objective (RPO) describes acceptable data loss, expressed as a period of time. Both require agreement from the business and validation by those responsible for the technology.

A four-hour target is not evidence that recovery can be achieved in four hours. A useful test record identifies what was restored, the conditions, the elapsed time and whether the business owner accepted the result. The target also needs to leave enough time, within the maximum disruption the business can tolerate, to validate the information and resume the work.

An application could meet its restoration target yet leave several days of accumulated transactions to process. That backlog is part of the business impact. So are dependencies that a supplier cannot restore to the same timetable.

Recovery targets need agreed starting and finishing points. Does the clock start when an outage occurs, when someone detects it or when a supplier accepts the case? Does 'restored' mean that a server is running, staff can log in or orders can be processed and verified? Ambiguity allows a provider to report technical success while the business is still unable to operate. A representative exercise should involve the service owner in accepting the business result, including data reconciliation and any backlog. An unsuccessful result is still valuable evidence if it leads to a better decision.

Use the map to direct effort

Often the first important finding is a shared dependency: a single identity platform, one specialist administrator or one critical supplier supporting several essential services. Addressing that concentration may matter far more than tackling a long list of isolated minor weaknesses.

I would keep the map as a working record, updated when a service changes, a supplier is replaced or an incident challenges an assumption. Its purpose is to help the business make decisions, not to produce a perfect diagram.

The first improvement identified by the map might be quite modest: retaining an accessible copy of essential delivery information, adding a second authorised administrator, or clarifying who may initiate recovery. These actions can be more useful than a large technology purchase when they remove the dependency most likely to frustrate the business. The map must also change as the service changes. A new supplier, additional integration or more demanding customer commitment may invalidate assumptions that were perfectly reasonable when the original document was created.

Sources and further reading

[S2] NCSC. Small Business Guide Response and Recovery. Step 1 Prepare for incidents. Read original guidance

[S3] NCSC. Introduction to identity and access management. Read original guidance

The source collection records these references as checked on 28 October 2026. Source attribution does not imply endorsement.

Scenarios are illustrative, not client case studies. This article provides general business guidance, not a security audit, certification or substitute for sector-specific technical, legal or contractual advice.