GUIDE · CONTINUITY AND DISASTER RECOVERY

Operational runbooks: how to write them so they work in a critical moment

After this guide you will know what makes a runbook usable during an emergency, how to write one step by step, how to test it with a person who did not write it and how to keep it up to date. The document must be executable by someone under pressure.

Updated october 2026
8 min read
Altovar Technical Editorial Team

1. Why a runbook matters in a critical moment

A runbook is an operating procedure written to carry out a specific task, such as restarting a service, restoring a database or isolating a compromised computer. In a critical moment the person using it is under pressure, may not be the one who performed the task last time, and must make quick decisions. A good runbook reduces choices to those that are necessary and reduces errors, because the reader does not have to reconstruct the sequence from scratch.

The runbook is different from the business continuity plan and the disaster recovery plan. These define what to do and who decides; the runbook defines how to carry out a single step. Without runbooks, the plan remains a list of intentions. With runbooks, each step becomes verifiable and repeatable by several people, including someone who has never seen the system in operation and must reach an answer without help.

2. Prerequisites

Some conditions are needed before writing a runbook. The task must be clear and bounded: a runbook that covers the entire recovery of the company is not usable. You need the person who really knows the procedure, because writing without observing it produces wrong or incomplete steps. And you need a storage location accessible even in an emergency, with an offline copy that does not depend on the internal network or a single computer.

Finally, check access. A runbook that requires a personal account of someone who is unavailable, or a password kept in a system that cannot be reached, is of no use at that moment. Emergency credentials must be managed separately, with a documented and controlled access procedure. The runbook says where to find them and who can authorise their use, without containing them. This way the document can circulate without exposing secrets.

3. How to write it: the steps

Write the runbook by observing a real or simulated execution, then reduce everything to verifiable steps. This sequence produces a uniform document, which is easier to follow under pressure. The first step is always to state what the document is for, because without a clear purpose the reader cannot tell whether it is the right one. Every subsequent step builds on this and must remain consistent with it.

Use direct language and the same format for all the company's runbooks. The reader should not have to interpret: they must carry out the action, check the result and move to the next step. Avoid generic phrases such as “if necessary”, which say nothing in an emergency, and replace them with observable conditions. For example, instead of “check the connection”, write what should appear on screen and what to do if it does not.

  1. Define the purpose in one sentence and the expected result at the end
  2. State when to use it and when not to, with the signals that trigger it
  3. List the prerequisites: access, tools and people to notify
  4. Write each step as a single action, with the command or screen to use
  5. Add for each step the expected result and what to do if it is not obtained
  6. Flag the decision points where confirmation or authorisation is needed
  7. Close with the final checks and the communication to be made

4. Common mistakes

The first mistake is writing a runbook that is too generic, such as “check the service”. The reader does not know what to check or how. The second is taking for granted what the reader does not know: a menu path, a server name, an internal abbreviation. The third is not stating what to do when a step fails, which is precisely the situation the runbook exists for. A runbook that describes only the happy path has no value at the critical moment.

Another frequent mistake is keeping the runbook in a place that will not be reachable during the emergency, or letting only the person who wrote it update it. When that person changes role, the document becomes a memory and nobody knows whether it is still valid. Duplicate versions also create confusion, because in an emergency it is unclear which copy to follow.

5. How to check that it works

A runbook is verified by running it. The best test is to give it to a competent person who did not write it, in a test environment, and observe where they stop or hesitate. Every question the person asks is a point to clarify in the document, and it must be rewritten until the question is no longer needed. This test is more useful than any review by someone who already knows the procedure, because that person takes for granted the same things the real reader does not know.

Also measure the execution time and compare it with the objective of the linked process. A runbook that is correct but slower than the objective needs rethinking: often checks can be brought forward or access prepared in advance, so that the time in an emergency is reduced. Record the measured time on the runbook card, together with the date of the test. The time should be compared with the objective at every review, not just the first time.

6. What to do after the first version

Every runbook should be linked to a review date and an owner. Changes to systems, such as an update or a change of supplier, must trigger the review of the affected runbooks. After each real incident, check which steps worked and which did not, and update the document with what emerged, without waiting for the planned periodic review. The most urgent corrections are those that cost time during a real event.

Each runbook card should show at least the following elements, so that whoever finds it during an emergency immediately knows whether the document is reliable, when it was last verified and whom to ask about a doubt. It is best to place them in the first part of the document, before any operational step, because they are the information the reader looks for first. They are few fields, but they make a difference:

  • Date of last verification and its outcome
  • Document owner and their substitute
  • Version and change log
  • Link to the process and the reference plan
  • Location of the offline copy

7. How to start

This week, choose the procedure your team carries out most often under pressure, for example restarting an email service or emergency access to a system. Ask the person who performs it to show you while they do it, and note every step, including the ones that seem obvious. The things that are obvious to the person performing the procedure are often exactly the ones missing from the document.

Then write the first version following the structure in section three, have a colleague read it and record the questions they ask. That first runbook will become the template for the others, and its test will give you a measure of the work to be done. Keep an offline copy from this first version onwards. A copy forgotten on your own computer is no use when that computer is out of action.

Frequently Asked Questions

What is the difference between a runbook and a playbook?

The term runbook usually refers to an operating procedure for a repeatable technical task, such as restarting a service. A playbook is more often used for incident response scenarios, with several phases, decisions and actors involved. In practice the two terms overlap, and the clarity of the text matters more than the name you give it.

Can a runbook be automated?

Some steps can be automated, but automation should be introduced with caution. An automatic step that fails silently is more dangerous than a manual one. Before automating, the manual runbook must have been verified successfully. Automation must then be tested like any other change.

How long should a runbook be?

As long as it takes to be executed without questions by a competent person. A simple task can fit on one page, while a more complex one can be divided into phases. Length is not an indicator of quality: a short and precise runbook is better than a long and ambiguous one.

Who should write the runbooks?

The person who carries out the procedure, under the supervision of a technical manager who checks accuracy and safety. The IT supplier can contribute for the systems it manages. The final text must be verified by someone other than the author, because that person spots logical jumps and steps taken for granted.

How do you protect runbooks that contain sensitive details?

Do not include passwords or keys in the text. The runbook indicates where to find emergency credentials and who can authorise their use, but the credentials remain in separate, controlled custody. Limit access to the document to those who need it, and log who consults it where the system allows.

About this content

General informational guide. Regulations and technical requirements change over time: always check the official sources and assess your specific case with a qualified technical or legal advisor.

GET IN TOUCH

Talk to our team.

We'll show how Altovar integrates into your infrastructure. No commitment, no sales pitch. Just a technical conversation.

BOOK A DEMO →