News · 19 March 2026
Eighty patches, two zero-days: counting vulnerabilities is not prioritising them
Microsoft's March 2026 Patch Tuesday, released on 10 March, fixed around eighty vulnerabilities, including a small number that were publicly disclosed or exploited before the fix. As usual, security teams were left to decide what to deploy first, on how many machines and how quickly.
Patch Tuesday is a good lens for a wider problem. Many organisations track "number of open vulnerabilities" as a headline measure of security. It is easy to count and poorly connected to risk.
Why counts mislead
A severity score such as CVSS describes how bad a flaw could be in general. It does not know whether the vulnerable feature is exposed in your environment, whether a working exploit exists or what sits behind the system. Two vulnerabilities with the same score can differ enormously in urgency.
A better prioritisation question set
- Is it being exploited? Public catalogues of known exploited vulnerabilities, and the Exploit Prediction Scoring System (EPSS), estimate this and are far more predictive than severity alone.
- Is it reachable? A flaw in a service that is not exposed to untrusted input is less urgent than one on an internet-facing system.
- What is the blast radius? A compromised build server or identity system is far more damaging than a kiosk.
- Do compensating controls exist? Segmentation, application firewalls or disabled features can reduce urgency, but they should be verified, not assumed.
- How hard is the fix? Sometimes the quick mitigation is the right first step.
The role of validation
Automated scanning is excellent at finding candidates. Deciding which are real, reachable and consequential needs context, and that is where human validation matters. In our approach, findings are checked by a security specialist and reported with evidence and an impact statement, so the output is a short, ordered list instead of a spreadsheet of thousands.
Metrics worth tracking
- Time to remediate exploited vulnerabilities on exposed systems.
- Coverage: the share of assets in your inventory that are actually scanned.
- Recurrence: how often the same class of issue returns.
These measure whether the programme is working. Raw counts mostly measure how many scanners you own.
Written in March 2026; details reflect the situation at that time.