Securafy | Knowledge Hub

NVD Enrichment Changes: What SMBs Should Do About CVEs

Written by Rodney Hall | Nov 9, 2026, 1:00:00 PM

If your vulnerability scanner says you have no critical issues this month, how sure are you that it is seeing every new vulnerability? Since April 2026 that question has a less comfortable answer. NIST changed how the National Vulnerability Database handles new CVEs: it now enriches, meaning adds severity scores and affected-product data to, only a prioritized subset, and labels the rest "Lowest Priority - not scheduled for immediate enrichment," according to NIST's announcement of the NVD changes.

The short answer for SMBs: the NVD enrichment changes mean many new CVEs arrive without a NIST severity score or product mapping, so tools that rely on NVD alone can under-rank or miss them. The fix is to stop treating one CVSS number as the patch trigger and to prioritize with exploitation data, vendor advisories, and an accurate asset inventory.

What exactly did NIST change in the NVD?

On April 15, 2026, NIST replaced its goal of analyzing every CVE with a risk-based model. It now prioritizes three groups for enrichment: CVEs in CISA's Known Exploited Vulnerabilities (KEV) catalog, which it aims to enrich within one business day; CVEs in software used by the federal government; and CVEs in "critical software" as defined by Executive Order 14028. Everything else is still published in the NVD, but not scheduled for NIST analysis unless someone requests it by email.

Three other changes came with it. NIST will no longer routinely add its own severity score when the CVE Numbering Authority (CNA) that submitted the CVE has already provided one. It will only reanalyze modified CVEs when a change materially affects the enrichment data. And it moved all backlogged, unenriched CVEs published before March 1, 2026, into a "Not Scheduled" category. The stated reason is volume: CVE submissions rose 263% between 2020 and 2025, and even a record year of nearly 42,000 enriched CVEs in 2025 could not keep pace.

Has anything changed since April?

Yes, and the follow-on changes matter to anyone reading NVD data. The NVD news page records that on April 28, 2026, NIST corrected about 4,500 CVE records, roughly 19% of those with a CVSS v4.0 score, whose numerical scores had been calculated too high. On June 17, 2026, the NVD began including Stakeholder-Specific Vulnerability Categorization (SSVC) data from CISA's Authorized Data Publisher, plus the "affected" product information from the CVE record itself, in its feeds and API. In late August it began adding both to the CVE detail pages.

NIST also signaled that more change is coming. Its request for information on modernizing the NVD, published August 12, 2026, asked how AI-enabled automation, better prioritization data, and interoperability could reshape the database, with comments due October 13, 2026. The direction is clear even if the end state is not: the NVD is becoming one input among several, not a complete, scored record of every flaw.

The real problem is not the missing score

Most coverage of this change focuses on CVSS scores. For an SMB, the bigger risk is the missing product mapping. Many scanners decide whether a CVE affects you by matching the product identifiers (CPE data) NIST adds during enrichment against the software they find on your machines. The Cloud Security Alliance's research note on the NVD overhaul warns that missing CPE data can make a CVE operationally invisible to tools that work this way, and estimates that roughly 80% of CVEs will be catalogued without NIST enrichment.

An unscored vulnerability at least shows up as a question. An unmapped one may never show up at all. That is why "our scanner shows nothing critical" is a weaker statement in late 2026 than it was two years ago, and why the question to ask your IT provider is where its tools get product and severity data when NIST has not supplied it.

There is a second uncomfortable point. CVSS was never designed to tell you what to patch first. It measures how bad a flaw could be in theory, not whether anyone is exploiting it. The NVD enrichment changes simply expose a prioritization habit that was already shaky.

What does the data show since the cutback?

Early analysis confirms the gap is real. Dark Reading reported on research by Volerion covering the first two months after the change, April 15 to June 15. Of 13,441 CVEs published to the NVD in that period, 8,342 were prioritized for enrichment, 6,759 were actually enriched, and only 2,645 received a NIST CVSS vector. In other words, about one in five new CVEs in that window carried NIST's own severity assessment.

The same research found scoring disagreements that should give anyone pause about treating a single number as truth. In one example, a denial-of-service flaw in IBM HTTP Server (CVE-2026-8856) was rated 9.1 critical by NIST, 7.7 by IBM, and 4.4 by Volerion. When CNAs supply the only score, those scores also vary in quality and incentive, since a vendor may understate a flaw in its own product while a researcher may overstate one they discovered.

Why does this matter if my MSP already manages patching?

It matters because patch automation inherits whatever data it is fed. If your provider's patch policy says "deploy critical CVSS fixes within seven days, everything else monthly," then a severe but unscored flaw quietly falls into the monthly bucket. Nothing fails loudly. The risk simply moves from your provider's dashboard into your environment.

There is also a documentation issue. Cyber insurers and auditors increasingly ask how you prioritize vulnerabilities, and many written policies still name CVSS from the NVD as the basis. If that source no longer covers most new CVEs, the policy no longer describes what actually happens. We cover what reviewers look for in what cyber insurers and auditors actually want to see from your security program.

This is also where the NVD change meets federal patching policy. CISA's risk-based directive for federal agencies, which we explained in what BOD 26-04 means for small business IT, prioritizes by active exploitation and internet exposure rather than raw severity. The NVD now enriches KEV entries first for the same reason. Both point SMBs toward the same model.

What should SMBs do about the NVD enrichment changes?

The goal is a prioritization process that does not depend on any single database being complete. For most 20 to 300 person companies, that comes down to five steps:

  1. Ask where your scanner's data comes from. Find out whether your vulnerability tool relies on NVD CPE matching or uses vendor and commercial feeds to identify affected products, and how it flags CVEs that have no score.
  2. Make KEV a standing trigger. Any CVE in CISA's Known Exploited Vulnerabilities catalog that matches your software should move to the front of the queue regardless of its CVSS score.
  3. Add exploit probability. FIRST's Exploit Prediction Scoring System (EPSS) publishes a daily probability that a CVE will be exploited in the next 30 days. It is free, and it helps rank the large middle of the list that KEV does not cover.
  4. Watch vendor advisories for your edge systems. For firewalls, VPNs, remote access tools, and line-of-business applications, the vendor's own security advisory is often faster and more precise than any database.
  5. Keep a versioned asset inventory. None of this works if you do not know exactly which products and versions you run. Inventory is what turns a CVE list into a patch list.

Then update your written vulnerability management policy so it names these sources rather than "NVD CVSS." That one edit keeps your documentation honest for the next insurance application or audit.

Is this the same as exposure management?

It is a practical starting point for it. Exposure management adds business context and control validation on top of exploitation data, which we walk through in how SMBs should prioritize security fixes with CTEM and exposure management. If you have already started there, the NVD change is a reason to check that your data inputs still hold up. If you have not, KEV, EPSS, and an inventory are the minimum. Unsupported software, the subject of our piece on technical debt as a cybersecurity risk, deserves extra attention, because it may never receive a fix at all.

Software your team adopts on its own adds to the inventory problem, including AI tools and add-ons that rarely appear in a scanner's view. Our AI Lab looked at that in whether your AI plugins were ever scanned before installation.

How Securafy handles vulnerability prioritization now

We do not treat a missing NVD score as "low." Our vulnerability prioritization checks active exploitation through the KEV catalog first, weights exploit likelihood and internet exposure next, and uses vendor advisories for the edge and remote access systems attackers target most. Every client environment is tied to a versioned inventory, so a new advisory can be matched to real devices in hours rather than waiting for a database to catch up.

We also write that logic into each client's documented vulnerability management process, so what appears on an insurance application matches what our team actually does. When a flaw is actively exploited on an internet-facing system, it goes through an emergency path rather than the monthly cycle.

Where To Go From Here

The NVD enrichment changes do not make vulnerability management impossible, but they do end the assumption that one public database scores everything for you. Start by asking your IT provider where its scanner gets product and severity data when NIST has not supplied it.

To see how prioritized vulnerability management fits into a broader prevention-first program, review Securafy's managed security services.

If you would rather talk through your specific environment first, book a strategy call with Securafy and we will walk through how your current patch prioritization handles unscored and unmapped CVEs.