
IT documentation: what to record before someone leaves
IT documentation is the job that waits until it is too late. While your IT person is around, you can just ask; the day they are off sick, on holiday or gone, you find out what only existed in their head. This is what goes in it, where it should live and who keeps it up to date.
The real risk is not the resignation but the absence nobody announces. Could someone else get into the firewall, renew the domain and restore a backup tomorrow?
Good IT documentation is not a manual but a fixed set of parts: network diagram, IP plan, admin rights, a password vault, licences, suppliers, backups and procedures.
It does not sit on the system it describes, it has an owner, and it is only finished once someone else has carried out a task using it.
The bus factor of one IT person
The bus factor is the number of people who have to disappear before something stops working. In many small and mid-sized organisations it is exactly one for IT: one colleague who knows how the network fits together, where the admin passwords are and which registrar holds the domain.
The dangerous moment is rarely the announced departure. A notice period gives you weeks to hand over. What hurts is the absence nobody sees coming: illness, an accident, a holiday without signal on the day the firewall refuses an update.
So a useful test for the board is not a question about documentation but about today: if our IT person is unreachable for the next two weeks, who can recover an admin account, renew a certificate and restore a deleted folder? If the answer is “nobody” or “I don’t know”, that is the brief.
What belongs in IT documentation
IT documentation does not need to be a thick manual. It is a fixed set of parts, each as short as it can be without anyone having to guess.
- Network diagram. Which sites, which connections, where the firewall, switches and access points sit and how they connect. A labelled sketch is enough, as long as it is right.
- IP plan. Which ranges and VLANs exist, what has a fixed address (printers, servers, cameras) and where DHCP runs. Without it, you rediscover the plan by watching devices drop off.
- Accounts and admin rights. Who holds which admin role in Microsoft 365, on the firewall, at the domain registrar and in the backup platform. Including the accounts that belong to no person.
- Licences and contracts. What you use, in whose name, when it expires and what the notice period is. Point to your IT asset register here rather than keeping a second copy.
- Suppliers and contacts. Internet, telephony, printers, software, the upkeep of the server room: who to call, with which customer number, and what the contract covers.
- Backup configuration and restore test. What is backed up, how often, to where, how long it is kept, and when something was last restored.
- Procedures. The tasks that come up rarely, so nobody knows them by heart: setting up a starter, closing down a leaver, renewing the domain, replacing a certificate, starting up after a power cut.
Passwords belong in a vault, not a spreadsheet
The password spreadsheet is the most common form of IT documentation and the worst. Anyone who can open the file has everything, you cannot see who looked at it, and the copy on someone’s desktop has been out of date for months.
A password vault solves three things at once. You share per vault instead of all or nothing, you can see who opened what, and when someone leaves you revoke access without having to change every password that same afternoon. Which vault suits you is covered in our piece on password management and passkeys.
Two traps we see often. The first is the admin account in the IT person’s own name, with multi-factor authentication on their personal phone: the access leaves with the phone. The second is the missing break-glass account: one admin account tied to no person, with a long password stored safely in two places, whose use gets noticed.
A backup without a restore test is an assumption
Almost everyone knows a backup is running. What is rarely written down is the part you need on the day it goes wrong: where the copy is, who has access to it, and how you bring back one file, one mailbox or a whole server.
So record the restore test, not just the backup. The date, what was restored, how long it took and what did not go to plan. A “successful” status in the backup platform tells you something was written, not that it can be read back.
This is also what the frameworks ask for. NIS2 names backup management and asset management among its measures, and ISO/IEC 27001 expects documented operating procedures. In the UK, the NCSC’s Cyber Essentials and 10 Steps guidance push in the same direction. If you write it down for yourself, you have it for the auditor as well.
Where it lives: not on the system it describes
The most common mistake is also the most logical one: the documentation sits on the file server, or in SharePoint. Until the file server is encrypted or the Microsoft 365 account will not open. Then the recovery guide is locked behind the very thing that needs recovering.
- Keep a copy outside the environment itself, for example in the password vault, which runs as a service separate from your own systems, or with a second party.
- Make sure at least two people can get to it, one of whom is not the IT person. Often that is a director, who does not need to understand it but does need to be able to open it.
- Keep the essentials offline as well: the break-glass account, the internet provider’s support number, the steps to open the vault itself. An envelope in a safe is old-fashioned and it works.
Who keeps it current, and how you know it is right
Documentation written once is out of date within a year. What works is not a big clean-up but a habit: a change is not finished until the documentation is updated. New switch, new supplier, new admin account: the ticket only closes once it is in the plan.
Give it an owner. That is usually the IT person, but the board should know that and ask once a quarter when it was last checked.
The real test is doing, not reading. Have someone else carry out a procedure with only the documentation in hand: create a user, restore a file. Wherever they get stuck, a line is missing. One such exercise a year tells you more than a document that looks complete.
In the weeks before someone leaves
When the person who carried IT is leaving, the notice period is the last chance to get what is in their head onto paper. Use those weeks for handover, not for wrapping up projects.
- Have the successor do the work while the leaver watches, not the other way round. What the leaver demonstrates looks clear; what the successor has to do themselves shows what is missing.
- Move every account held in their name: the registrar, supplier portals, app store accounts, the multi-factor authentication. That is separate from closing their own account, which comes afterwards.
- Ask about what is written down nowhere: the yearly task, the printer that only works after a restart, the supplier you have to ring rather than email.
If there is nobody to hand over to, that is the moment to decide whether the knowledge goes back to one new colleague or to a team.
Where our work stops
What we do. For an environment we manage, we write and maintain the IT documentation as part of the job: network diagram, IP plan, rights, suppliers, backups and procedures. When we take over an environment, it starts with mapping what is there, and from then on the bus factor is a team rather than one person.
The documentation and the vault are yours, not ours. They sit in your environment or in a vault you own, with access for someone from your own organisation. A managed service provider who is the only one with the passwords is the bus factor of one in a different coat.
What we do not do. We do not write business processes outside IT or a quality manual for the whole organisation. We do not review contracts legally: we record what is running and when it ends, but what a clause means is for a lawyer. And we certify nobody: whether documentation meets ISO/IEC 27001 is decided by the certification body.
Questions we get about this
The five that come up most.
What should IT documentation include?
At minimum a network diagram, an IP plan, an overview of accounts and admin rights, a password vault, licences and contracts with their end dates, suppliers with contacts, the backup configuration with the last restore test, and procedures for the tasks that rarely come up. Short is fine, as long as nobody has to guess.
Where should IT documentation be stored?
Not on the system it describes. If the recovery guide sits on the file server that has just been encrypted, it is no use to you. Keep a copy outside your own environment, make sure two people can reach it, and keep the break-glass account and the steps to open the vault offline too.
Should passwords be kept in the documentation?
Not in a document or a spreadsheet. Passwords belong in a password vault, where you share per section, see who opened what and can revoke access. The documentation points to the vault; it does not contain the passwords.
How often should IT documentation be updated?
With every change, as a fixed part of that change, plus a short check each quarter. The best test is having someone else carry out a procedure once a year using only the documentation.
Do you write our IT documentation, and who owns it?
For an environment we manage, we write and maintain the documentation. It stays yours: it sits in your environment or in a vault you own. We do not write business processes outside IT, we do not review contracts legally and we certify nobody.
Where this sits with us
The parts that come together in your documentation.
Shall we look at how big your bus factor is?
Tell us how your IT is run today and who knows what. We will show you which parts of the documentation exist, which are missing, and what has to come first if your only IT person is unreachable tomorrow.
Practical IT knowledge in your inbox
New guides on management, security and the workplace, written by the people doing the work. No sales talk, and you can unsubscribe in one click.