# A scoring rule is an ordered list: KEV first, then EPSS and asset criticality, with CVSS as the tie-breaker rather than the gate

RISK SCORING, read 10 Nov 2026

**The short answer.** Write the rule as an ordered ladder. A KEV listing sets the top tier. Otherwise EPSS and asset criticality decide, and CVSS breaks ties. Against 318 KEV additions, a CVSS filter of 7 or higher misses 33 (10.4 percent), and an EPSS cutoff of 0.1 misses 197 of 260 scored (75.8 percent) on listing day.

## Three inputs and what each one answers

A usable rule combines three questions. The first is whether the vulnerability is being exploited now. That is what the [CISA Known Exploited Vulnerabilities catalog](https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json) records, with a due date per entry. The second is how likely exploitation is soon. That is [EPSS](https://www.first.org/epss/), which FIRST describes as a data-driven model estimating the probability that a published CVE will be exploited in the wild in the next 30 days, published daily as a 0 to 1 value with a percentile. The third is how much it matters here. That is your asset criticality, which no public feed can supply.

CVSS answers a fourth question, how severe the flaw is in the abstract, and [FIRST's CVSS pages](https://www.first.org/cvss/) say as much. Keep it in the rule as a modifier. The numbers below show why it should not be the gate.

## What the KEV Lag Ledger says about each input

The KEV Lag Ledger on this publication covers every catalog addition from 2025-11-10 to 2026-11-09: 318 entries, each joined to its NVD record and, where one existed, its EPSS score on the day it was listed. Formula and sources are in the ledger's method note: days from NVD publication to listing, CVSS base score from NVD (v3.1 first), EPSS from the FIRST API for the listing date.

- CVSS: 33 of 318 entries (10.4 percent) have a base score below 7. A rule that only escalates CVSS 7 and above leaves those out until someone notices. 146 (45.9 percent) are 9.0 or above.
- EPSS: of the 260 entries with a score on listing day, 197 (75.8 percent) were below 0.1, and the median was 0.0066. A rule that only escalates EPSS 0.1 and above would not have flagged most of them on the day CISA did.
- Timing: the median gap from NVD publication to listing is 10.5 days. 144 entries (45.3 percent) were listed within 7 days, and 55 (17.3 percent) were more than a year old at listing.
- Ransomware: 29 entries are marked as known to be used in ransomware campaigns in the catalog.

## The ladder

Here is a worked rule. Tiers are labeled P1 to P4, and the SLA column is an illustration, not a recommendation; the SLA guide covers how to set the clocks.

Asset criticality is a four-level field (Critical, High, Medium, Low) populated from the inventory. Rules are evaluated top to bottom and the first match applies, which is the same pattern ServiceNow documents for its assignment rules ([ServiceNow docs](https://www.servicenow.com/docs/api/khub/maps/x8~ohSTeWLioGLVI5x3Vgg/topics/CYDyzm_AEkiaQuEnwDs3Ng/content)).

| Order | Condition | Tier | Illustrative SLA |
| --- | --- | --- | --- |
| 1 | In KEV catalog, any asset | P1 | KEV due date, or 7 days if tighter |
| 2 | Not in KEV, EPSS 0.1 or higher, asset Critical or High | P2 | 14 days |
| 3 | Not in KEV, EPSS 0.1 or higher, asset Medium or Low | P3 | 30 days |
| 4 | Not in KEV, EPSS below 0.1, CVSS 9.0 or higher, asset Critical | P3 | 30 days |
| 5 | Everything else | P4 | 90 days, ordered by CVSS then criticality |

## Two real entries run through the rule

CVE-2021-26829, an OpenPLC ScadaBR cross-site scripting flaw, was added to KEV on 2025-11-28. NVD scores it CVSS 5.4, and its EPSS on that day was 0.00252. A CVSS gate and an EPSS gate would both have passed over it. Under the ladder, it is P1 on any asset the moment it is listed, because rule 1 does not look at the other fields.

CVE-2020-7796, a Synacor Zimbra Collaboration Suite server-side request forgery, was added on 2026-02-17 with CVSS 9.8 and EPSS 0.90602 on listing day. Before listing, on a Critical asset it would already have matched rule 2 (P2); on listing it moves to P1. The ladder does two jobs: it escalates what the catalog confirms, and it ranks the rest by likelihood and exposure while waiting for the catalog.

## Details that decide whether the rule holds up

Record the evaluation time and the input values with each decision. EPSS changes daily, so a finding's tier on Monday may differ from Tuesday; auditors ask why a finding moved. Keep a changelog of the rule itself, with date and approver.

State the thresholds as parameters, not constants. The 0.1 above is an example. FIRST's FAQ notes that exploitation has a low base rate and that roughly 2.5 to 3 percent of published CVEs show exploitation activity in a 30-day window, so any cutoff trades coverage against workload; test candidates against your own backlog and pick the one your teams can absorb.

Decide what happens to findings with a missing input. No EPSS score (58 of the 318 ledger entries had none on listing day) should default to a tier, not to nothing. The same goes for assets with no criticality: default to High until classified, so gaps are visible rather than comfortable.

## Test the rule before you automate it

Replay the last 90 days of findings through the ladder and count how many land in each tier. If P1 and P2 together exceed what the teams close in a month, tighten rule 2 or widen the SLA. If P1 is empty, check that the KEV feed is connected and refreshing daily. Then compare against what the program actually fixed first. Large gaps point to either a rule problem or an ownership problem.

Once the ladder is stable, configure it in the platform. Brinqa documents customer-set risk factors on top of a base score and ServiceNow documents customer-weighted calculators; the rule above should be expressible in either. If a candidate platform cannot express an ordered rule with KEV as a hard override, note it in the proof of concept.

## Sources

- CISA, Known Exploited Vulnerabilities catalog (JSON feed): https://www.cisa.gov/sites/default/files/feeds/known_exploited_vulnerabilities.json (read 10 Nov 2026)
- FIRST, Exploit Prediction Scoring System: https://www.first.org/epss/ (read 10 Nov 2026)
- FIRST, EPSS frequently asked questions: https://www.first.org/epss/faq (read 10 Nov 2026)
- FIRST, Common Vulnerability Scoring System: https://www.first.org/cvss/ (read 10 Nov 2026)
- ServiceNow documentation, assignment rules: https://www.servicenow.com/docs/api/khub/maps/x8~ohSTeWLioGLVI5x3Vgg/topics/CYDyzm_AEkiaQuEnwDs3Ng/content (read 10 Nov 2026)

Last reviewed 10 Nov 2026. Canonical: https://riskbasedvulnerabilitymanagement.com/guides/how-to-write-a-risk-scoring-rule/
