Vulnerability prioritization: rank patches by CVSS, EPSS or CISA KEV, and run it in-house or with an outside provider?
Quick answer, Nov 20, 2026As of 20 November 2026, no single score should rank your patches. FIRST says CVSS Base scores "are not risk". EPSS estimates exploitation in the next 30 days. KEV records exploitation already seen. CISA's BOD 26-04 ranks by exposure, KEV status, automation and technical impact. A provider can scan; the decision not to fix stays yours.
Four signals, four different questions
| Signal | What it answers | What it does not | Where it is written |
|---|---|---|---|
| CVSS Base score (CVSS-B) | How severe the flaw is, usually as scored by the product vendor or a bulletin analyst | Your risk: "CVSS-B Base scores are not risk, and should not be used alone for patch prioritization" | CVSS v4.0 FAQ, "How does CVSS provide input to patch priority?"; CVSS v4.0 specification, "Assessment" |
| EPSS | "the probability that a published CVE will be exploited in the wild in the next 30 days" | Impact or your estate: "EPSS is not a complete risk score" | FIRST EPSS; EPSS FAQ, "What is EPSS not?" |
| CISA KEV | "reliable evidence that the vulnerability has been actively exploited in the wild" | What comes next: KEV and similar sources "only report exploitation that have already occurred" | CISA KEV criteria, "How should organizations use the KEV catalog?"; NCSC principle 2, "Monitoring vulnerability feeds", Note |
| SSVC | A decision (Track, Track*, Attend or Act) from five values, including exploitation status and automatable | A fit for every firm: CISA's tree is built for government and critical infrastructure, and SEI publishes other trees | CISA SSVC, opening section, "How CISA Uses SSVC" and "Additional SSVC Decision Tree Models" |
When KEV and EPSS disagree, FIRST's own answer under "My vulnerability is on CISA KEV but has a low EPSS score. Which should I follow?" is two words: "Follow KEV." KEV vs EPSS covers reading the two together.
A ranking you can copy: BOD 26-04's four questions
BOD 26-04, issued 10 June 2026, binds federal civilian executive branch agencies. Its "Background" section sets urgency from four questions: is the asset publicly exposed, is the CVE in the KEV catalog, can an adversary automate the exploit, and does exploitation give partial or total control. The same section says it "supersedes and hereby revokes" BOD 22-01, so a policy that still cites BOD 22-01 timelines cites a revoked directive.
Appendix A, Table 1 turns the four answers into deadlines. Row 1 (exposed, in KEV, automatable, total control) reads "3 days & forensic triage". Row 8 (exposed, not in KEV, not automatable, partial control) reads 60 days. Row 16 (no to all three, partial control) reads "Fix on system upgrade".
Neither a CVSS score nor an EPSS score appears in the table. CISA's implementation guidance, under "What is the point of creating a new updating requirement? Should my agency still use CVSS for prioritization?", says "the FCEB no longer requires CVSS use for vulnerability prioritization." CVSS has not vanished, though. The same answer says the directive "aligns with" FIRST's CVSS consumer guidance, and Appendix A calls technical impact "similar to" the CVSS base score's idea of severity.
You are probably not bound. At the top of CISA's KEV guidance page, CISA says every organization, "Although not bound by BOD 26-04", can strengthen its posture by prioritizing KEV entries. FIRST's CVSS FAQ calls the directive "a model for other organizations to use or adapt".
Our reading: CISA publishes the KEV, automation and impact answers for every CVE ("Background"). Exposure is the one answer CISA cannot give you. Footnote 2 under Table 1 says it may come from "third-party asset management or vulnerability management services or scanners", but it is only as good as your asset list.
When there is no patch yet
NIST SP 800-40r4 (April 2022), section 3.3 "Define Risk Response Scenarios", names "Emergency mitigation" as a procedure "to temporarily mitigate vulnerabilities before a patch is available". Section 3.5.3 adds: "Schedules should be set and enforced for both patch deployment and mitigation removal." The NCSC, principle 4 "The triage process", says a temporary workaround "should be time-bounded and actively tracked".
Who decides
The NCSC's principle 5, The organisation must own the risks of not updating, is direct. Under "Thinking behind this principle": "The decision not to fix an issue is, at root, a senior-level business risk decision, not an IT problem." Under "Things to consider": "You shouldn't make decisions based purely on a single severity score, such as CVSS," and "the business owns the risk, not the security team".
SP 800-40r4 section 2.1 "Risk Responses" lists four answers to a vulnerability: accept, mitigate, transfer or avoid. NIST CSF 2.0 asks that "Risk responses are chosen, prioritized, planned, tracked, and communicated" (ID.RA-06), with roles and authorities established (GV.RR-02).
In-house or an outside provider
No source we opened names one right model. The split below is our reading of what each source assigns.
| Task | Who can run it | Where it is written |
|---|---|---|
| Asset inventory | You own the list of what you run | SP 800-40r4 section 2.2, step 1 |
| The exposure answer | Either: a scanner or provider can supply it from your list | BOD 26-04, Appendix A, Table 1 footnote 2 |
| Scanning | Either. It "doesn't need to be run by an external partner" | NCSC principle 4, "Scanning" |
| Watching vulnerability feeds | Either: "the general vulnerability management process or a procured service" | NCSC principle 2, "Monitoring vulnerability feeds" |
| Applying updates | Usually "an IT admin or operations team (rather than the security team)" | NCSC principle 2, "Integrate with IT outage / incident policies" |
| Deciding not to fix | You, at senior level | NCSC principle 5, "Thinking behind this principle" |
If you contract any of it, write the speed into the contract. The NCSC, principle 2 "Identify teams / individuals", says critical suppliers and managed service providers should be "contractually liable to rapidly mitigate vulnerabilities in internet-accessible systems" when those are exploited in the wild. CISA's implementation guidance, under "Vendors, Third Party Vendors, and Cloud Service Providers", recommends that federal agencies using outside providers ensure "appropriate remediation timelines are part of a service-level agreement". CSF 2.0 GV.SC-05 puts supply chain requirements into "contracts and other types of agreements". When you choose a scanning service, ask how fast it detects a newly disclosed flaw. The NCSC's Vulnerability scanning tools and services, "Additional considerations", puts detection at "no more than a few days for critical issues."
If the same provider would also watch your systems around the clock, MDR, an in-house SOC or co-managed covers that choice. If one KEV listing is already live in your estate, start with the first 24 hours of an exploited VPN flaw. If a vendor's own systems were breached, see the website script vendor guide.
From an AI assistant, get_cve on the Distributed Risk MCP server returns a CVE's record, KEV listing and EPSS estimate, each dated. To scope an assessment or a managed service, brief an advisor and we arrange a partner firm.
Questions
Should we multiply CVSS by EPSS? FIRST's EPSS FAQ, "How is EPSS different from CVSS?", says doing so "is never a good idea", because the result "is no longer a probability".
What EPSS threshold should we use? The same FAQ, "What threshold should I use?": "There is no universal threshold." It adds that 0.10 "is commonly cited in the field but carries no special authority from EPSS."
Does BOD 26-04 apply to us? Only to federal civilian executive branch agencies. Its "Scope" section says that, unless a procurement contract directs it, it "does not apply to contractors". No ranking method guarantees you will not be breached.
Is this legal advice? No. This page is a procedural guide drawn from the public record.