Patch Management Policy

How to prioritize patches

Prioritize a patch by first choosing a risk response, then setting speed. CISA's BOD 26-04 (issued 10 Jun 2026) sets urgency from four variables: asset exposure, Known Exploited Vulnerabilities status, exploit automation and technical impact. Its fastest tier is three days plus forensic triage. Read 19 Nov 2026.

Choose the response before the speed

NIST SP 800-40 Rev. 4 section 3.3 lists four responses: accept (rely on existing controls), mitigate (including patching, but also disabling a feature or adding segmentation), transfer (insurance, or moving to SaaS where the provider patches) and avoid (uninstall or decommission). SP 800-218 asks for the same decision: “Make a risk-based decision as to whether each vulnerability will be remediated or if the risk will be addressed through other means.” A patch is one of several outcomes of this step, and the outcome is written down before the clock starts. Choosing a mitigation does not close the item: see exceptions clause for the record.

The numbers each framework allows are on the patch window table, and the policy wording is in severity and patch windows. The process overview shows where this step sits.

The four variables CISA uses

Source: CISA BOD 26-04, read 19 Nov 2026. The directive binds federal civilian agencies; other organizations may borrow the method.
VariableQuestion in BOD 26-04Who supplies the answer
Asset exposureIs the vulnerable asset publicly exposed?The agency, using CISA's Internet Exposure Reduction Guidance
KEV statusIs the CVE ID on CISA's Known Exploited Vulnerabilities catalog?CISA, for every CVE
Exploit automationCan an adversary automate all the steps needed to exploit it?CISA's Vulnrichment Program
Technical impactDoes the adversary gain partial or total control after exploitation?CISA's Vulnrichment Program

The directive's Table 1 turns these four answers into deadlines. The page text states one value outright: the entries marked “& forensic triage” must be remediated or mitigated within three days and the asset triaged for compromise. Other tiers are given in the table figure, which this site read by eye and does not reproduce here; see the directive for them. Days are calendar days.

What changes mid-clock

BOD 26-04 treats each asset's timeline as “dynamic as facts change”: removing a system from the internet changes “Publicly Exposed” from Yes to No and moves the required timeline back, while a vulnerability newly added to the KEV catalog shortens it. The timeline starts at whichever comes first, the catalog addition or the agency identifying the vulnerability on an asset. A private policy can copy this by stating that the clock is per asset and restarts on a stated change of exposure, with the change recorded.

Why speed matters: one dated measurement

Mandiant's analysis of 138 vulnerabilities exploited in the wild and disclosed in 2023 found an average time to exploit of five days, down from 32 days across 2021 and 2022 and 63 days across 2018 and 2019 (published 15 Oct 2024; it excludes 15 outliers, and without that exclusion the 2023 average is 47 days). It is a third-party analysis of one year's sample (Google Cloud, Mandiant). On 18 Nov 2026 the KEV feed held 1,786 entries, 361 of them flagged as used in known ransomware campaigns (counts from the CISA feed).

The record to keep

  • The risk response chosen per release and per maintenance group, with the date and the approver.
  • The four variable answers used, and where each came from.
  • The deadline derived, and the date it was revised if exposure changed.
  • For emergencies, the scenario used; see emergency patch clause.

Sources

Last reviewed 19 Nov 2026