Back to blog
Pentesting

Vulnerability management: which CVEs to fix first and when a pentest adds value

Sectricity Security TeamNovember 19, 2026

More than 40,000 CVEs a year and no time to patch them all. How to rank vulnerabilities with CVSS, EPSS and the CISA KEV catalogue, what NIS2 expects and when a pentest shows what a scanner misses.

Vulnerability ManagementCVEPentestingNIS2Patch ManagementRetest
Person working on laptops with program code on the screens

TL;DR

  • More than 40,000 CVEs were published in 2024. Nobody patches all of them, so the order in which you fix them matters more than the total.
  • Combine three signals: CVSS for severity, EPSS for the likelihood of exploitation and the CISA Known Exploited Vulnerabilities (KEV) catalogue for flaws that attackers already use.
  • Add your own context: is the system reachable from the internet, and what does an attacker get if it falls?
  • A scanner finds known CVEs. A pentest shows which ones are really exploitable, how they chain together and which flaws no scanner knows.

Every week brings a new batch of vulnerability reports, and every scanner turns them into a long list sorted by CVSS score. In 2024 more than 40,000 CVEs were published, according to an analysis of the CVE data by Jerry Gamblin. No IT team can treat that as a to-do list.

The good news: only a small share of those vulnerabilities is ever used in real attacks. A study by the Cyentia Institute put that at about 6% of all published CVEs. Good vulnerability management means finding that share in your own environment first. This post explains how to rank CVEs, how fast to act and where a penetration test adds something a scanner cannot.

Why a CVSS score alone puts the wrong things first

The Common Vulnerability Scoring System, maintained by FIRST, gives each vulnerability a score from 0 to 10. Version 4.0 has been available since 1 November 2023. The ratings are fixed: 0.1 to 3.9 is low, 4.0 to 6.9 medium, 7.0 to 8.9 high and 9.0 to 10.0 critical.

That score describes how bad a vulnerability is under general assumptions. It says nothing about whether anyone is exploiting it, or about where the system sits in your network. If you sort by CVSS alone, a critical flaw in a library on an internal test server ends up above a high flaw in the firewall that faces the internet. The first may never be touched. The second may already be in an attacker's toolkit.

Three signals that tell you what to fix first

CVSS: how bad could it be?

Use the CVSS score as a first filter on impact. A flaw that allows remote code execution without authentication deserves attention, whatever else is true. But treat the score as a property of the vulnerability, not as your risk.

EPSS: how likely is exploitation?

The Exploit Prediction Scoring System, also maintained by FIRST, estimates the probability that a published CVE will be exploited in the wild in the next 30 days. The score updates daily. A CVE with a modest CVSS score but a high EPSS score often belongs higher on your list than its severity suggests.

KEV: is it already being exploited?

The CISA Known Exploited Vulnerabilities catalogue lists vulnerabilities with confirmed exploitation in the wild. CISA calls it the authoritative source for that question and recommends that all organisations use it as an input for prioritisation. If a CVE on your systems appears in the KEV list, the discussion about probability is over.

In Europe, ENISA launched the European Vulnerability Database in May 2025. It brings vulnerability information from several sources together and is a useful second reference next to the CVE and KEV data.

Add your own context

The three signals describe the vulnerability. Two questions about your own environment decide the final order:

  • Exposure: can an attacker reach the system from the internet, or only from inside the network?
  • Impact: what does an attacker get if it succeeds? Full control of a domain controller weighs more than read access to a brochure site.

In June 2026 CISA replaced its old directive for US federal agencies with Binding Operational Directive 26-04. It ranks vulnerabilities on four variables: public exposure, presence in the KEV catalogue, whether exploitation can be automated, and whether it gives total or partial control. The directive only binds US federal agencies, but the model is a good template for any company.

A practical order of work

  1. KEV vulnerabilities on systems reachable from the internet: fix or isolate them first, within days.
  2. KEV vulnerabilities on internal systems, and flaws with a high EPSS score on internet-facing systems.
  3. Critical and high CVSS scores on business-critical systems.
  4. Everything else in your normal patch cycle, with a written reason when you decide not to fix something.

Write that order down as a short policy with target times per group, and keep a record of when each fix was done. That record is what an auditor or a customer's security questionnaire will ask for.

What NIS2 expects

Article 21(2)(e) of NIS2 names vulnerability handling and disclosure among the measures that essential and important entities must take. The directive does not prescribe a scanner, a frequency or a deadline. It expects a process that you can show: how you learn about vulnerabilities, how you decide what comes first, and how you verify the fix. See our NIS2 compliance page for the wider picture.

What a scanner sees and what a pentest adds

A vulnerability scanner compares version numbers and configurations with a database of known CVEs. That is fast and repeatable, and you should run it on a fixed schedule. Our posts on pentesting versus vulnerability scanning and what a vulnerability assessment is go deeper into that difference.

A scanner cannot tell you three things:

  • Exploitability in your setup. A CVE may be present but unreachable because of how the system is built, or reachable in a way the scanner did not test.
  • Chaining. Three medium findings can together lead to full control of the network. A scanner lists them separately and ranks none of them first.
  • Flaws without a CVE. Broken access control, business logic errors in a web application and weak configurations in cloud identities never get a CVE number, so no scanner database contains them.

An ethical hacker tests those questions by hand. The result is a short list of findings that are proven to work, ranked by what they mean for your business. Our guide on how to read a pentest report explains how to turn that list into actions.

When a pentest adds the most value

  • Before a new application, API or customer portal goes live.
  • After a major change, such as a migration to the cloud or a new remote access solution.
  • When a KEV vulnerability hits your perimeter, to check whether the patch closed the door and whether anyone came in earlier. Our post on critical PHP flaws shows what that looks like in practice.
  • When a customer, an insurer or an auditor asks for proof that your measures work.

For companies that need a test at short notice or several times a year, RedSOC offers pentesting on demand: you book a test when a change or a new CVE calls for it.

Frequently Asked Questions

Which CVEs should I fix first?

Start with vulnerabilities that are listed in the CISA Known Exploited Vulnerabilities catalogue and sit on systems reachable from the internet. Then look at flaws with a high EPSS score and a high CVSS score on important systems. A critical CVSS score on an isolated test server can wait longer than a medium score on your VPN gateway that attackers already use.

What is the difference between CVSS and EPSS?

CVSS describes how severe a vulnerability is if someone exploits it, on a scale from 0 to 10. EPSS estimates the probability that a published CVE will be exploited in the wild in the next 30 days. CVSS answers how bad it could be, EPSS answers how likely it is. You need both, plus context about your own systems.

What is the CISA Known Exploited Vulnerabilities catalogue?

It is a public list maintained by the US Cybersecurity and Infrastructure Security Agency of vulnerabilities that have been exploited in the wild. US federal agencies must fix them within set deadlines. CISA recommends that every organisation uses the list as an input for its own prioritisation, and that is useful advice for European companies too.

Does NIS2 require vulnerability management?

Yes. Article 21(2)(e) of NIS2 lists security in the acquisition, development and maintenance of systems, including vulnerability handling and disclosure, among the measures that essential and important entities must take. The directive does not prescribe a tool or a deadline, so you decide the process and you have to be able to show that it works.

Is a vulnerability scan enough, or do I need a pentest?

A scan is the right tool to find known CVEs and missing patches, and you should run it regularly. A pentest shows whether those findings can really be exploited, how they combine into an attack path and which flaws no scanner recognises, such as broken access control or business logic errors. Most companies need both, at different frequencies.

How do I prove that a vulnerability is really fixed?

Rescan the system, and for the findings that matter most, have them tested again by hand. At Sectricity a retest of the findings is included in every pentest when it is done within 90 days of the report, so you get evidence that the fix works and not just that a patch was installed.

Related services and resources

Sectricity's ethical hackers run penetration tests on web applications, networks and cloud environments, with a retest of the findings included within 90 days of the report. For tests at short notice, RedSOC gives you pentesting on demand. If you are not sure where to start, request a free security scan of your external exposure.