Patch Management Policy

How to write a patch management policy

Write a patch management policy in nine sections, in this order: purpose, roles, inventory, windows, testing, maintenance windows, exceptions, reporting, review. Fix the windows last, from the frameworks that bind you, as read on 19 Nov 2026: Cyber Essentials v3.3 asks for 14 days, BOD 26-04 for 3, 14 or 60.

Steps

  1. List the frameworks that bind you. Each one decides which numbers you may not exceed; start from the patch window table.
  2. Group the assets. NIST SP 800-40 Rev. 4 section 3.4 defines a maintenance group as assets that "generally have the same software maintenance needs"; a group, not a server, gets a rule.
  3. Write a plan per group for each scenario NIST names: routine, emergency patching, emergency mitigation, unpatchable assets (section 3.5).
  4. Set windows by severity, naming the clock start (release, discovery, or KEV listing).
  5. Add a canary stage to routine plans: NIST section 3.5.1 has a small subset receive the patch first, expanding if impact is minimal.
  6. Write the exceptions clause from the exceptions clause.
  7. State the metrics you will report; see patch compliance rate.
  8. Name an approver and a review interval. CIS Controls v8 safeguard 7.1 uses annual review.
  9. Attach, do not embed, the product list and tool settings.

Skeleton to fill

Patch Management Policy

1 Purpose and scope

This policy applies to [asset classes] owned or managed by [organization].

2 Roles

Approver: [role]. Deployer: [role]. Reporter: [role].

3 Inventory

No asset is in scope of a patch deadline until it is in [inventory name].

4 Windows

Critical: [days] from [release | discovery | KEV listing].

High: [days] from [clock start].

Other: as set in the maintenance plan for the group.

5 Testing

Routine patches go first to [canary group] for [days].

6 Maintenance windows

[Day and hour] per group; reboots are [rule].

7 Exceptions

See the exceptions clause. Each exception has an owner, a reason, a control and an end date.

8 Reporting

[Measure] is reported to [role] every [period].

9 Review

Reviewed every [12 months] or after [event]. Approved by [role] on [date].

The fully worded versions are in the template set. Windows left in brackets are yours to set: NIST, NIS2, DORA and NYDFS 500.5 give no number, so the table cannot supply one.

Order matters

Writing windows first produces numbers nobody can meet. Writing the group definitions and the exceptions clause first shows which assets cannot meet any window, and those become the exceptions rather than a silent miss. The process steps show where each section is used.

4.3 Patch windows

SeverityWindow3143090180 days
1. Known exploited, internet-facing3 calendar daysApplies to: Systems reachable from the internet
1. Known exploited, other14 calendar daysApplies to: All other systems
2. Critical14 calendar days
3. High30 calendar daysApplies to: All systems [set 14 if you hold Cyber Essentials]
4. Medium90 calendar days
5. Low180 calendar days
Routine baselineEvery month, or more oftenApplies to: Operating system and application updates

No framework selected: every window above is the site template's suggested default, and the clock starts as this section states.

Questions

how to create a patch management policy

Write nine sections in order (purpose, roles, inventory, windows, testing, maintenance windows, exceptions, reporting, review) and set windows last, from the frameworks that bind you. Group assets first, as NIST SP 800-40 Rev. 4 does with maintenance groups.

Sources

Last reviewed 19 Nov 2026