How much backup coverage capacity goes untested?

Question and scope

This review asks how much of a role's critical work has a backup who has been tested. The unit is a critical task that must continue during an absence. The scope covers assistant roles with recurring client deadlines and named coverage. The review does not estimate a universal coverage rate, because the critical work and the acceptable delay vary by role.

The question matters because a backup plan often exists only on paper. A named person may lack the access, the instructions, or the authority to complete the work. The primary assistant may also be the only person who knows where the files live. A plan that has never been tested is a hope rather than a control.

The review separates coverage capacity from coverage intent. Intent is the named backup in the plan. Capacity is the tested ability to complete the work within the service standard. A role can have a named backup and still have no real coverage. The review measures the capacity, not the name.

Select the critical tasks

List the tasks that cannot wait for the primary to return. Include the daily client update, the time-sensitive approval, the payment request, and the access change. Ask the client which tasks would cause a problem if they waited two days. The answer may differ from the internal list.

Rank the tasks by the cost of a delay. A missed regulatory filing is different from a delayed internal report. Use the ranking to decide which tasks need a tested backup and which can wait. Not every task needs the same coverage. The ranking is a decision about where to spend the testing effort. A short list of truly critical tasks is easier to test and maintain.

Write the service standard for each critical task. The standard should say how quickly the work must be completed and what evidence shows it was done. A backup who can do the task in twice the time may still fail the standard. The standard is part of the coverage definition.

Test the three layers of coverage

The first layer is access. The backup needs a named account, the right permission, and a working second factor. Test the login and the specific action, not only the account status. A backup who can open the tool but cannot approve the request is not covered.

The second layer is instruction. The backup needs a current runbook that they can find without the primary. Test the search path from the place the backup actually starts. A document in the primary's personal folder is not a runbook. The instructions should name the source, the steps, and the escalation contact.

The third layer is authority. The backup needs permission to make the decisions the task requires. A backup who must ask the primary for every approval is not available when the primary is absent. Name the decisions the backup may make and the ones that need another owner. Record the authority in the plan.

Measure the result

For each critical task, record whether the backup completed it, the time taken, the questions asked, and the errors found. A completion without the standard is not full coverage. Report the partial cases separately. The pattern shows which layer is missing.

Calculate the share of critical tasks with tested coverage and the share with a named but untested backup. Add the share with no backup at all. The three numbers tell a more useful story than a single coverage percentage. A team may have names for every task and tests for none.

Compare the drill time with the normal time. A backup who takes much longer may still be acceptable for a low-frequency task but not for a daily one. Use the service standard to decide. The comparison also shows where the runbook or the tool is slowing the backup down.

Interpret the gaps

An access gap is usually the easiest to fix, but it can also be the most sensitive. Grant the minimum permission for the task and remove it after the drill. Do not leave permanent broad access just to make the backup plan look complete. The temporary grant should have an owner and an expiry.

An instruction gap points to the runbook or the search path. A backup who could not find the document may still know the work from experience. The test should use a person who does not have that experience, or the result will overstate the coverage. Write the runbook for the person who was not there.

An authority gap is a management decision. The backup may need approval authority for a limited set of actions during an absence. Write the delegation, the limit, and the end date. A verbal permission is hard to prove when a client or an auditor asks who approved the work.

Controls that help

Run a drill on a fixed cadence and record the result. Start with one critical task and one named backup. Test the access, the instruction, and the authority in the same exercise. A small, regular drill is more useful than a large annual rehearsal.

Keep the runbook in the shared workspace and link it from the task. Review it after every drill and after any change in the client or the tool. Assign an owner and a review date. A runbook without a date is assumed to be current even when it is not.

Give the backup a way to escalate a gap without blame. The drill should surface problems, not punish the person who finds them. Record the gap and the repair. The next drill should test the repair, not repeat the same discovery.

Method and limitations

This review describes a local coverage assessment rather than a study of backup capacity across providers. The sample depends on the tasks the team identifies and the people available to test them. A drill with a knowledgeable backup may overstate the coverage. A drill with an inexperienced backup may understate it.

The drill can also change behavior. People may prepare more carefully because they know the test is coming. That is a useful effect, but it makes the first result a weak baseline. Repeat the drill and use the later result for comparison. Note the date of the first test.

Decision checklist

List the critical tasks and rank them by the cost of a delay. Write the service standard and the evidence of completion. Name the backup and the decisions they may make. Record the date the list was last reviewed.

Test access, instruction, and authority in one drill. Record completion, time, questions, and errors. Report tested, named-but-untested, and no-backup shares. Repair the missing layer and test again.