Request a quote
Emergency SectorsCareers About us Blog Get in touch
Close-up of two monitors showing log lines and coloured terminal output

Patch management: who updates what, and when

Patch management sounds like a single switch in Windows Update. In practice it is a schedule with owners: who updates which device, in what order, in which window, and what happens when an update breaks something. The laptops are usually sorted. The firewall, the printer and that one server in the cupboard often are not.

6 min read Written by the ITproposal team
Short answer

Laptops are rarely the problem. What falls behind are the things without automatic updates: firewall and switch firmware, printers, hypervisors and software that does not come from Microsoft.

Roll out in rings: a small test group first, then everyone else, with a fixed wait in between. That catches the update that breaks something before the whole organisation has it.

A flaw that is already being exploited does not wait for the monthly window. Agree in advance who decides then, and what happens to devices that no longer receive updates.

Patch management is a schedule, not a switch

Almost every organisation has automatic updates switched on for its laptops, and so it feels handled. What is missing is everything around it: a list of what needs updating, an order, an agreed time, and someone who checks that it actually worked.

Patch management answers four questions for each type of device: who updates it, how often, in which window, and how you know it happened. Where one of those four reads “it happens by itself”, that is usually the device still on the same version a year from now.

It helps to treat it as an agreement rather than a technical project. The tooling (Intune, a management platform, the firewall’s admin console) carries out what has been agreed. Without an agreement it carries out whatever the default setting happens to be.

Everything that needs patching

Start with the list, because you cannot update what you do not know you have. These are the groups, roughly from well handled to poorly handled in a typical environment:

  • Workplace operating systems. Windows and macOS. Usually managed through Intune or a similar platform, and this is where patching tends to go best.
  • Office and the Microsoft 365 apps. They update themselves through their own channel. The choice is which channel: monthly, or the semi-annual one for organisations that want fewer changes.
  • Browsers. Edge and Chrome update themselves, but only once the browser restarts. A browser left open for three weeks runs an old version for three weeks.
  • Third-party software. PDF readers, Java, Zoom, the accounting client, tools someone once installed themselves. This is the largest gap, because no operating system updates them for you.
  • Servers. Updating is easy; the restart has to be planned. That second part is why it slips.
  • Hypervisors and storage. The layer beneath your servers. Updated least often, and it takes everything down at once if it goes wrong.
  • Network and peripheral firmware. Firewalls, switches, access points, printers and multifunction devices. The firewall sits on the outside edge, and it is the one most often forgotten.

Give each group an owner. Not a department, but a name or a supplier. That one column turns an inventory into a plan.

Rings: a few first, then everyone

An update that breaks something on a hundred laptops at once is an outage. The same update on five laptops is a ticket. That is why you roll out in rings.

  • Ring 0: IT itself. The administrators get the update first, on the day.
  • Ring 1: a pilot group. A handful of users from different teams, running the applications that matter: finance, the scheduling system, the till software. A few days later.
  • Ring 2: everyone. A few days after that, if nothing has come back from ring 1.

Microsoft releases its security updates on the second Tuesday of the month, which gives you a fixed rhythm to build the rings around. In Intune these are called update rings; Windows Autopatch does the grouping for you. If you still run WSUS: Microsoft is no longer developing it, so this is a moment to move rather than rebuild.

The trap is a pilot group that tests nothing. If ring 1 only contains colleagues who send email and browse, nobody notices that the update breaks the scheduling system. Choose the pilot group by what they use, not by who complains least.

Maintenance windows, and who agrees them

Laptops can update themselves against a deadline: the user may postpone, but after a few days the restart goes ahead. That does not work for servers, firewalls and hypervisors. They need a window in which a restart affects nobody.

Agree that window once, and write down who may move it. For example: the third Wednesday of every month, in the evening, servers and network. A window that has to be requested every month gets skipped once a month because “this week isn’t quite convenient”.

For each system, ask two things: what stops working while it restarts, and is there a way back if the update fails. For a server that means a snapshot or backup just before the window. For a firewall it means the previous firmware and the configuration, kept separately. Without a way back, every window is a gamble, and gambles get postponed.

When a flaw is already being exploited

The monthly rhythm suits routine work. It does not suit a vulnerability that attackers are known to be using already. Waiting for the next window is then exactly the risk you set out to avoid.

Define in advance what counts as an emergency patch, and who may push it through. A practical rule: an update goes outside the window if the vulnerability is actively exploited and the system is reachable from the internet. Firewalls, VPN appliances, mail servers and web applications go first. Advisories from the NCSC and the vendor’s own notices are the signals to watch.

In the UK there is a concrete benchmark: Cyber Essentials expects critical and high-risk updates to be applied within fourteen days of release. That is a ceiling for routine updates, not a target for an active exploit.

Also agree what happens if there is no patch yet. Sometimes there is only a temporary measure: switch off a feature, close a port, restrict access to a fixed address. That is not a fix; it buys time, and it belongs in the same log as the update that follows.

Devices that no longer get updates

Every device reaches the day the manufacturer stops. Windows 10 received its last regular security update in October 2025; Office 2016 and 2019 stopped at the same time. For switches, printers and firewalls that date is somewhere on the manufacturer’s website, and rarely anyone looks.

Put that end-of-support date in your inventory, next to the device. Replacement then becomes a line in next year’s budget rather than a surprise. What we regularly see: a network switch that works perfectly and has had no firmware for years, or a PC that only drives an old measuring instrument or machine.

If something cannot go, there are three things to do: put it in its own network segment, give it no internet access, and record it as a consciously accepted risk, with a name and a date to look at it again. That last step is not paperwork. It is the difference between an exception and something forgotten.

Showing that it happens

Patching without reporting is patching on trust. A good monthly report answers three questions for each group in the list above: how many devices are up to date, which are not and why, and how long the oldest outstanding update has been waiting.

That last figure is the most useful. A total that says “nearly everything” hides the one server that is four months behind. The list of exceptions, with reason and owner, says more than the total.

You need that report for another reason too. NIS2 names vulnerability handling explicitly as a measure, and an auditor will not ask whether you patch but how you show it. Cyber insurers often ask a similar question in their proposal form. A report that is produced every month as a matter of course is the answer you already have.

Where our work stops

What we do. When we manage the workplaces and the network, we build the inventory, set up the rings and windows, update operating systems, Microsoft 365 apps, browsers, third-party software, servers and network firmware, and deliver the monthly report with the exceptions in it. When a flaw is actively exploited, we act on the agreement made beforehand. That is part of our managed IT.

What we don’t do. Software from another supplier, such as an ERP system, a clinical system or the controls of a machine, is updated by that supplier or according to its instructions; we plan the window with them and make sure the network around it is right. The decision to leave a system unpatched on purpose is yours, not ours: we set out the risk and record it. And we are not an insurer and not an auditor: we provide the reporting a questionnaire or audit asks for, but we do not sign that questionnaire and we do not certify anyone as compliant.

Frequently asked

Questions we get about this

The five that come up most often.

What is patch management?

Patch management is the fixed arrangement for updating everything that runs software: workplaces, servers, applications and network firmware. It comes down to four agreements per type of device: who updates it, how often, in which window, and how you check it worked. Automatic updates on laptops are part of it, not the whole of it.

How often should you patch?

For workplaces and servers a monthly rhythm is usual, because Microsoft releases its security updates on the second Tuesday of the month. Firewall and switch firmware follows the manufacturer’s schedule. In the UK, Cyber Essentials expects critical and high-risk updates within fourteen days. A flaw that is actively exploited goes outside the window, especially on systems reachable from the internet.

What is the difference between patch management and vulnerability management?

Vulnerability management finds the weak spots, for example with a vulnerability scan. Patch management closes them, where an update exists. The two belong together: the scan shows whether patching works, and it also finds devices that sit in no update cycle at all.

What do you do with devices that no longer get updates?

Replacing them is the best answer, which is why the manufacturer’s end-of-support date belongs in your inventory. If a device cannot go, put it in its own network segment without internet access and record it as a consciously accepted risk, with an owner and a date to review it.

Will you complete our cyber insurance questionnaire?

No. That questionnaire is a statement from your organisation to the insurer, and the person responsible signs it. We do provide the facts it asks for: which devices are up to date, within what timeframe, and which exceptions exist. We are not an insurer and not an auditor, and we do not certify anyone as compliant.

Related services

Further reading

What sits around it: the management, the scan and what you must be able to show.

Shall we look at what is falling behind?

Tell us what you run and who updates it today. We will tell you which groups are covered, which are not, and which devices no longer receive updates.

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.

We use your address for the newsletter only. See our privacy policy.

Request a quote Call