Vulnerability Scanning: How It Works and What to Fix First
Skip to content

Vulnerability Scanning

Vulnerability scanning is the automated examination of systems for known weaknesses, such as missing patches, outdated software and insecure configuration.

On this page

A vulnerability scanner identifies each host and the software on it, then compares the result with a database of known vulnerabilities. A scan reports weaknesses that are probably present, and each result needs a person to interpret it.

This guide covers how a scanner works, the types of scanner and scan, frequency and coverage, prioritizing results, the path from scan to fix, and how a scanning program is measured, staffed and started.

How a Vulnerability Scanner Works

A vulnerability scanner finds hosts, identifies the software running on them and matches what it finds against known vulnerabilities. NIST defines vulnerability scanning in its glossary as a technique used to identify hosts, host attributes and associated vulnerabilities.

The NIST Technical Guide to Information Security Testing and Assessment (SP 800-115) describes the work, which most scanners carry out in this order.

  1. Discovery. The scanner sends probes across an address range and records the hosts that respond. Only hosts that are on and connected during the scan are found.
  2. Port and service identification. It checks each host for open ports and identifies the service listening on each, such as a web server or a remote access service.
  3. Fingerprinting. It works out the operating system, and the product and version behind each service. Banner grabbing, a common method, reads the information a service transmits when a connection opens.
  4. Vulnerability checks. It matches the operating systems and applications it identified against its vulnerability database, which flags outdated versions, missing patches and misconfigurations. Some checks also send a request and compare the response with the signature of a known vulnerability.
  5. Reporting. It lists each finding against the host, with a severity level and guidance on mitigation.

The database decides what the scanner can recognize. NIST advises installing the latest database updates before running any scanner, and treating update frequency as a major consideration when selecting one.

That database is assembled from public vulnerability records and the scanner maker’s own checks.

  • CVE identifier: A publicly disclosed vulnerability is typically assigned a CVE ID, the common key that findings, advisories and catalogs share.
  • National Vulnerability Database: The NIST glossary describes the NVD as the US government repository of standards-based vulnerability management data, which includes software flaws, misconfigurations, product names and impact metrics.
  • Check: The scanner maker writes a check for each record, so a new vulnerability can be detected only once its check ships. The NCSC expects that to take a few days at most for critical issues.

A vulnerability assessment is the wider review that a scan feeds, adding validation, context and a ranked report.

What a Scan Finding Records

A scan finding records the affected asset, the weakness and the observation that led the scanner to report it. The observation lets the system owner check the finding before any work is scheduled.

Field What it holds How to check it
Asset and service The IP address, hostname, port and service where the weakness was seen. Match it to the asset inventory and name an owner.
Product and version The product name and version read from a banner, a file, a registry key or the package list. Compare it with the version installed on the host.
Vulnerability identifier The CVE ID and a reference to the vendor advisory. Read the advisory for the affected and fixed versions.
Detection output The response, file path or setting that triggered the check. Repeat the observation with a single request or command.
First seen and last seen The scans in which the finding first and most recently appeared. Use the dates to measure time to fix and to spot findings that return.

NIST notes that scanners can have a high false positive rate, and that someone with expertise in networking and operating system security should interpret the results. A finding that rests on a version string alone may be false where the vendor fixed the product without changing the reported version.

Types of Vulnerability Scanner

Vulnerability scanners are grouped by the kind of target they assess, and the broadest division is between infrastructure and applications. The NCSC’s guidance on choosing a scanner describes each type.

  • Infrastructure scanner: It finds hosts and network services, fingerprints them and compares them with products known to be vulnerable. Typical results are missing patches, unsupported software, default or weak passwords, cleartext services and weak encryption settings.
  • Web application scanner: It works through an application over HTTP much as a browser would, sending requests built to draw out a response that indicates a weakness. Typical results are injection flaws, broken authentication and vulnerable third-party components.
  • Native software scanner: It runs on the same host as the software under review, or with access to its source code, and belongs to the software development process.
  • Specialist scanner: Separate products assess cloud infrastructure, mobile applications, or web applications built on one platform.

The NCSC advises establishing general infrastructure scanning first, then adding specialist scanners where the organization exposes those kinds of asset. An infrastructure scanner suits an estate built mostly from off-the-shelf products, and a web application scanner matters most where custom applications make up most of the external footprint.

A web application scanner returns useful results only when it is tuned to the application. The NCSC points to a configured sign-in page with credentials and the ability to exclude pages or kinds of check, since an untuned scan may fill a database with entries from repeated form submissions.

Authenticated and Unauthenticated Scans

An authenticated scan signs in to each host and reads its software and configuration directly, while an unauthenticated scan sees only what the host shows to the network. The choice sets how much a scan can find.

  • Unauthenticated scan: The scanner holds no credentials and works from open ports, banners and responses, so it infers versions from the outside.
  • Authenticated scan: The scanner signs in with an administrator-level account and reads installed packages, patch levels and settings. It also finds weaknesses in software that listens on no port, such as browsers and PDF readers.
  • Agent-based scan: A scanner installed on the host reports from the inside, which typically suits laptops that are rarely on the office network during a scan.

Standards favor credentials for internal scanning. PCI DSS requirement 11.3.1.2 calls for internal scans to be authenticated with sufficient privileges, and for systems that cannot accept credentials to be documented. CISA’s Binding Operational Directive 23-01 requires federal agencies to run vulnerability enumeration on managed endpoints and network devices with privileged credentials, to the maximum extent possible.

The scan account is a privileged account and is protected as one, with any use from outside the scanner’s own addresses treated as suspect.

External and Internal Scans

An external scan runs from the internet against the addresses an organization exposes, and an internal scan runs from inside the network. NIST observes that internal scanning usually uncovers more vulnerabilities than external scanning, and that testing from both viewpoints is important.

  • External scan: The scanner reaches only the ports that perimeter devices allow through. The result is the view any outside party has.
  • Internal scan: The scanner sits behind the perimeter and reaches far more hosts and services. The result shows what an intruder with a foothold, or a malicious insider, could reach.

The external target list comes from discovery, which the guide to external attack surface management explains.

Some external scanning is available from government. CISA’s Cyber Hygiene services scan internet-accessible assets at no cost for US government bodies and critical infrastructure organizations, and deliver weekly reports.

How Often to Scan

Scan frequency follows the exposure of the asset and the standard the organization works to, and published expectations run from daily to once every three months.

Source Applies to What it expects
PCI DSS version 4, requirement 11.3 Organizations that handle payment card data. Internal and external scans at least once every three months and after any significant change. External scans are performed by an Approved Scanning Vendor.
ASD Essential Eight maturity model, Maturity Level One Organizations that follow the Essential Eight. A scanner used at least daily for online services and the operating systems of internet-facing servers and network devices, at least weekly for office suites, web browsers, email clients, PDF software and security products, and at least every two weeks for other operating systems.
CISA Binding Operational Directive 23-01 US federal civilian agencies. Automated asset discovery every 7 days and vulnerability enumeration across discovered assets every 14 days.
NCSC vulnerability scanning guidance Any organization. Infrastructure scans at least once every month, or immediately after changes made to fix a critical issue. Application scans whenever the application changes.

ASD’s newer framework, ASD Essentials, sets no fixed interval. Its assessment evidence guide rates scanning as fully effective when all applicable assets are included in routine scanning and the frequency enables a timely response.

The sources agree that internet-facing systems are scanned most often, and that a significant change or a reported fix calls for a scan outside the schedule. Frequency for penetration tests is covered in the guide to continuous penetration testing.

Checking Scan Coverage

A scan result is only as complete as its coverage, so the scan’s own records are checked before the findings are read. A short report may reflect a well-patched estate, or hosts that were off, sign-ins that failed and checks that were out of date.

CISA’s Binding Operational Directive 23-01 separates the activities involved. Asset discovery identifies which network-addressable assets reside on the network, and vulnerability enumeration identifies and reports suspected vulnerabilities on those assets. Coverage is how closely the second matches the first.

Coverage check What it reveals Where to look
Scanned assets against the inventory Inventory entries that no scan reached, and scanned hosts that the inventory lacks. The scanner’s asset export, joined to the asset inventory on hostname or address.
Credential check for each host Whether the sign-in succeeded, and with enough privilege to read packages and settings. The authentication results in the scan output, listed host by host.
Hosts that gave no answer Addresses in range that were off, asleep or filtered while the scan ran. The scan log, compared with last check-in times in endpoint management.
Signature update date The age of the vulnerability database that the scan used. The scan details, or the scanner’s update log.
Scan policy The ports and checks in force, and whether denial of service checks were suppressed. The policy or template export from the scanner.
Exclusion list Assets left out on purpose, each with a reason and a review date. The risk register.

Several of these records are requirements in their own right. BOD 23-01 requires detection signatures to be updated within 24 hours of the vendor’s last release, and it includes roaming devices in each enumeration cycle. PCI DSS requirement 11.3.1 calls for the scan tool to be kept up to date with the latest vulnerability information, and the NCSC asks for a record of every asset excluded from scanning.

Prioritizing Scan Results

Scan results are prioritized by combining severity, evidence of exploitation, likelihood of exploitation and the exposure of the affected asset. A scan typically returns more findings than a team can fix at once.

CVSS

The Common Vulnerability Scoring System (CVSS) is an open framework for communicating the characteristics and severity of software vulnerabilities, owned and managed by FIRST. A score runs from 0 to 10 and maps to a rating: low from 0.1 to 3.9, medium from 4.0 to 6.9, high from 7.0 to 8.9 and critical from 9.0 to 10.0.

In version 4.0, Base metrics describe qualities of the vulnerability that stay constant, and Environmental metrics are unique to the user’s environment. A score built from Base metrics alone, labeled CVSS-B, leaves out the system the vulnerability sits on.

Known Exploited Vulnerabilities Catalog

The CISA Known Exploited Vulnerabilities (KEV) catalog lists vulnerabilities that have been exploited in the wild. CISA adds an entry when these conditions are met.

  • Identifier: The vulnerability has an assigned CVE ID.
  • Exploitation: There is reliable evidence that it has been actively exploited.
  • Remediation: There is a clear remediation action, such as a vendor update.

Each entry shows whether the vulnerability is known to be used in ransomware campaigns. CISA strongly recommends that all organizations monitor the catalog and prioritize remediation of the vulnerabilities it lists.

EPSS

The Exploit Prediction Scoring System (EPSS), maintained by a special interest group at FIRST, estimates the probability that a published CVE will be exploited in the wild in the next 30 days. Scores run from 0 to 1 and are published daily. FIRST describes EPSS as the threat likelihood component of a risk calculation.

Signal Question it answers Where to get it
CVSS score and vector How severe is the weakness by its own characteristics? The scan finding and the vendor advisory.
KEV listing Has it been used in real attacks? The CISA catalog, matched on CVE ID.
EPSS probability How likely is exploitation in the next 30 days? The daily scores from FIRST, matched on CVE ID.
Asset exposure Can the affected service be reached from the internet? External scan results and the asset inventory.
Asset importance What does the system hold or control? The asset inventory and the system owner.

SSVC

Stakeholder-Specific Vulnerability Categorization (SSVC) turns those signals into a decision. CISA’s SSVC decision tree weighs exploitation status, technical impact, whether exploitation is automatable, mission prevalence and public well-being impact. Each vulnerability receives one of these outcomes.

  • Track: The vulnerability needs no action at this time, and the fix follows standard update timelines.
  • Track*: The vulnerability has characteristics that may need closer monitoring for changes, within the same timelines.
  • Attend: Supervisory-level staff give it attention, and the fix comes sooner than standard update timelines.
  • Act: Supervisory-level and leadership-level staff give it attention, and the fix comes as soon as possible.

What to Fix First

The findings to fix first are exploited vulnerabilities on exposed assets, followed by those most likely to be exploited next. CISA’s Binding Operational Directive 26-04, Prioritizing Security Updates Based on Risk, was issued in June 2026 for federal civilian agencies. It ranks a vulnerability by public exposure of the asset, KEV listing, how far an attack can be automated and how much control the attacker gains, and the guide to continuous threat exposure management sets those factors out.

The directive binds federal agencies only, and the same factors give other organizations a workable order.

  1. Known exploited, on an internet-facing asset. Fix or mitigate at once, and review the asset’s logs for earlier use. Where the directive calls for forensic triage, agencies must also assess whether the system is compromised.
  2. Known exploited, on an internal asset. Fix ahead of the routine cycle.
  3. Likely to be exploited, on an exposed asset. A high EPSS probability or a critical rating on a reachable service comes next.
  4. Severe, on an important internal system. Identity systems, management tools and data stores come before general workstations.
  5. The remainder. Fix in the routine patch cycle.

ASD gives a fixed point for the top of the list. At Maturity Level One of the Essential Eight, patches for online services are applied within 48 hours of release when the vendor rates a vulnerability as critical or a working exploit exists.

From Scan to Fix

A scan result becomes a fix when someone owns it, a response is chosen and a rescan confirms the outcome.

  1. Validate. Check the detection output against the host, and close false positives with the reason recorded.
  2. Assign. Give each finding to the owner of the asset, and group findings that one patch will close into a single task.
  3. Choose the response. NIST’s Guide to Enterprise Patch Management Planning (SP 800-40 Rev. 4) lists the responses as accept, mitigate, transfer and avoid. Mitigation covers patching, disabling a vulnerable feature and isolating the asset.
  4. Apply the change. Test the patch and deploy it through change management.
  5. Rescan. Run the same check against the asset. PCI DSS requires rescans that confirm high-risk and critical vulnerabilities have been resolved.
  6. Keep watching. NIST includes continuous monitoring of the response, so a patch that is later removed is caught.

A Finding Followed to Closure

A single finding followed through a week shows where the time goes and which records are kept. The system is a file transfer server on a public address, in an organization that scans external systems daily and internal systems weekly.

  1. Monday, detection. The external scan reports a vulnerability rated critical in the file transfer product, inferred from the version shown on its sign-in page. The first seen date is set.
  2. Monday, validation. The system owner reads the installed version on the server and compares it with the vendor advisory. The version falls in the affected range, so the finding is confirmed.
  3. Monday, ranking. The CVE is absent from the KEV catalog and its EPSS probability is high. As a likely target on an exposed asset, the server is booked for the Wednesday maintenance window, and its administration page is limited to the office network in the meantime.
  4. Tuesday, new evidence. CISA adds the CVE to the KEV catalog. The finding moves to the top of the order, and the patch is applied that evening as an emergency change.
  5. Tuesday, rescan. The same check runs against the server alone and returns clean. The finding is closed with a time to fix of one day.
  6. Wednesday, log review. The vulnerability was under exploitation while the server was exposed, so the access logs since disclosure are reviewed for the activity the advisory describes.
  7. The following Monday, coverage. The weekly internal scan reports the same finding on a standby server that had been switched on for a recovery test. It is patched before it is next brought into service.

The ranking changed within a day because the evidence changed, which is why open findings are ranked again at every scan and every catalog update.

Scanning Program Measures

A scanning program is measured on coverage and on speed of repair, broken down by the importance of the asset and of the vulnerability. In SP 800-40 Rev. 4, NIST treats organization-wide totals, such as the number of vulnerabilities found and the percentage patched, as too simple to act on.

  • Coverage: This is the share of the inventory scanned within its required interval.
  • Credentialed coverage: This is the share of scanned hosts where the sign-in succeeded.
  • Fixed by deadline: This is the percentage of findings closed inside the agreed deadline, for each pairing of asset importance and vulnerability severity.
  • Time to fix: This is the mean and median days from first seen to closed, for the same pairings.
  • Open known exploited findings: The report lists every open finding with a KEV listing, with its age and owner.
  • Returning findings: These are closed findings that reappear in a later scan.

NIST’s example reports these figures as a matrix, and adds that an organization scanning only monthly counts far fewer vulnerabilities than it has, and that unauthenticated scans generally underreport them.

First Weeks of a Scanning Program

A new scanning program starts with named roles, the target list and the external view, then adds credentials and widens.

  • Security lead: The lead decides the scope, the schedule and the ranking rule, and approves exclusions.
  • Scanner operator: The operator keeps targets, credentials and signatures current. PCI DSS expects qualified personnel who are organizationally independent, and its example is that a network administrator should not be responsible for scanning the network.
  • System owners: The owners supply maintenance windows, validate findings, apply fixes and request rescans.

In a smaller organization the same person often covers more than one role. An MSP or MSSP scanning for several clients holds each client’s scope, written authorization, scan account and results apart. With the roles named, the work runs in this order.

  1. Build the target list. Collect assets from the directory, endpoint management, DHCP leases, cloud accounts and public DNS, the records that describe the attack surface, and sort them into groups such as internet-facing systems, servers and workstations.
  2. Scan the external group. An unauthenticated external scan covers the systems that any outside party can reach.
  3. Prepare the scan account. Create a dedicated privileged account and confirm that the sign-in works on a sample of each platform.
  4. Trial the internal scan. Scan a small group first. For servers that host business-critical services, the NCSC suggests scanning non-production instances first, where their configuration is consistent with production.
  5. Widen to every group. Add groups until the inventory is covered. Where scanning everything is impractical, the NCSC advises giving priority to assets that are internet-accessible, host business-critical services or contain the most sensitive data.
  6. Agree the ranking rule and deadlines. The first full scan typically returns a long backlog, so system owners agree the order and a deadline for each level before tickets are raised.
  7. Fix the schedule. Set recurring scans for each group, a rescan on every reported fix and a monthly report of the measures.

Limits of Vulnerability Scanning

Scanning finds known weaknesses on the assets it is pointed at, and each of its limits has a practical answer.

  • Known vulnerabilities only: A weakness with no signature goes unreported, so vendor advisories for products in the inventory are followed as well.
  • Coverage follows the target list: An asset missing from the list is never scanned, so asset discovery runs first and the inventory is kept current.
  • Weaknesses in isolation: A scanner rates each weakness alone, and NIST notes that several low-risk vulnerabilities can present a higher risk when combined.
  • Application behavior: The NCSC notes that network scanners generally lack the ability to navigate web applications. Access control between accounts needs a web application scanner with credentials, or a test.
  • Load on fragile systems: Scanning adds network traffic and may affect older hosts. The NCSC allows a temporary exclusion recorded in the risk register, and treats the fragility as a vulnerability to fix.

Whether a reported weakness can be used is a question for a penetration test, and the guide to penetration testing vs vulnerability scanning sets out that comparison.

Vulnerability Scanning With ThreatDefence

A scan says a weakness is probably present, and ThreatDefence agentic penetration testing checks the most urgent of those findings for real exploitability. It runs inside an AI SOC platform with a 24x7 SOC behind it, where security monitoring, compliance and cyber risk management are automated together.

  • Advisory as trigger: A new advisory for a product the organization exposes starts a test, ahead of the next scheduled run.
  • Exposed targets: The ASM agent keeps the map of internet-facing hosts and services, so the affected systems are already known.
  • Confirmation: The testing agent tests them within the confirmed and authorized scope and confirms which weaknesses are exploitable.
  • Result: A confirmed finding carries proof, a severity based on exploitability and fix steps, and SOC analysts review significant findings.

Where the customer is also monitored, the Hunter and Defender agents check its telemetry for earlier use of the weakness. The organization’s own internal scanning continues alongside, and the agent retests when a fix is reported.

Frequently Asked Questions

What is a vulnerability scanner?

A vulnerability scanner is a tool that examines hosts for known weaknesses. It discovers the hosts in a range, identifies their open ports, services, operating systems and software versions, and matches them against a database of known vulnerabilities. Its report lists each finding with a severity level and a recommended fix, and the findings need review before they are acted on.

What is the difference between authenticated and unauthenticated scanning?

An unauthenticated scan has no credentials and judges a host from its open ports and responses. An authenticated scan signs in with a privileged account and reads the installed software, patch levels and configuration directly. Authenticated scans typically find more, including weaknesses in software that listens on no network port, and they produce fewer findings based on inference.

How often should vulnerability scans run?

It depends on exposure and on the applicable standard. PCI DSS requires internal and external scans at least once every three months and after significant changes. The NCSC advises scanning infrastructure at least once every month. ASD’s Essential Eight maturity model asks for at least daily scanning of online services and internet-facing systems.

Why do vulnerability scanners report false positives?

Scanners match signatures, and they often infer a vulnerability from a version number or banner without testing for it. A system with a backported fix or a non-default configuration can match the signature while being unaffected. NIST SP 800-115 notes that scanners can have a high false positive rate and that results need interpretation by someone with the right expertise.

Can a vulnerability scan disrupt production systems?

Yes, in some environments. NIST notes that vulnerability scanning generates significantly more network traffic than port scanning and may have a negative impact on the hosts or network being scanned. Scanners often allow denial of service checks to be suppressed. Scans of sensitive systems are usually run in agreed windows, at reduced rates, after a trial against a non-production copy.

What is the difference between CVSS and EPSS?

CVSS describes how severe a vulnerability is, on a scale from 0 to 10, based on its characteristics. EPSS estimates the probability, from 0 to 1, that a published CVE will be exploited in the wild in the next 30 days. FIRST, which maintains both, states that they measure different things. Prioritization typically uses the two together with the CISA KEV catalog.

Should every critical vulnerability be fixed first?

No. A critical CVSS rating describes the vulnerability and says little about the affected system. A lower-rated vulnerability that appears in the CISA KEV catalog and sits on an internet-facing asset usually comes ahead of a critical one on an isolated internal host. Exposure, evidence of exploitation and the importance of the asset set the order.

Can a vulnerability scanner find zero-day vulnerabilities?

No, in most cases. A scanner recognizes weaknesses that are already in its database, so a vulnerability with no public record has no check to match. A scan may still report conditions that make an unknown vulnerability easier to use, such as an exposed management interface or an unsupported product, and vendors typically add checks soon after a vulnerability is disclosed.

Who can perform vulnerability scans for PCI DSS?

External scans under requirement 11.3.2 are performed by a PCI SSC Approved Scanning Vendor at least once every three months. Internal scans, and scans after a significant change, can be run by qualified internal staff or a specialist firm, provided the person scanning is reasonably independent of the systems being scanned. A Qualified Security Assessor is not required for either.

Is a port scan the same as a vulnerability scan?

No. A port scan identifies the open ports on a host and the services listening on them. A vulnerability scan does the same and then attempts to identify weaknesses in what it found, by matching products and versions against a vulnerability database. Its output is a list of probable weaknesses, where a port scan returns a list of services.

Talk to Our Experts

See how ThreatDefence automates security monitoring, compliance and cyber risk management in one platform.

    I'm an MSP, MSSP, or reseller

    By submitting this form, you accept the terms in our privacy policy.