Vulnerability scan, security check or penetration test: what does your organisation need?
Key takeaways
- A vulnerability scan automatically finds known vulnerabilities and missing updates. It runs continuously and forms the foundation, but it does not recognise attack paths.
- A security check (security assessment) reviews the configuration against recognised baselines and delivers a prioritised list of actions. It is the right starting point if no structured review has taken place yet.
- A penetration test uses a controlled attack to show whether the measures hold against a real attacker.
- The three do not replace one another. Over a year they work together: scan continuously, review regularly, test in a targeted way, and evidence remediation.
Anyone who needs evidence of their security posture will be offered three terms that are often used interchangeably: vulnerability scan, security check and penetration test. For management the distinction is not a formality. The three cost different amounts, answer different questions and are weighted differently by auditors, customers and insurers. Commission the wrong one and you pay for an answer to a question you did not ask. This article sets out the three approaches from a penetration tester’s perspective and shows how they fit into an annual plan.
What distinguishes a scan, a security check and a pentest?
A vulnerability scan is an automated comparison. A tool checks systems against a database of known vulnerabilities (CVEs) and identifies missing updates, outdated software and common misconfigurations. In Microsoft environments this is often handled by Microsoft Defender Vulnerability Management: the core capabilities are included in Defender for Endpoint Plan 2, and for servers also in Defender for Servers. For the external view, external scanners check your public IP addresses and services.
A security check, which we call a security assessment, is an expert review of the configuration. It is based on recognised baselines such as the Microsoft cloud security benchmark, CIS Benchmarks or BSI IT-Grundschutz. It covers, for example, Conditional Access policies, role assignments, guest access, Intune device policies and logging. The result is a gap list with a priority and a specific action for each item.
A penetration test is an authorised, manual attack. Testers actually exploit weaknesses and chain them into attack paths. The guiding question is how far an attacker gets, and by which route.
| Vulnerability scan | Security check (security assessment) | Penetration test | |
|---|---|---|---|
| Objective | Find known vulnerabilities and missing updates | Assess configuration against baselines | Demonstrate exploitability and attack paths |
| Method | Automated, signature-based | Expert review of settings, roles and processes | Manual, controlled attack supported by tools |
| Depth | Broad but shallow | Broad, moderate depth | Targeted and deep |
| Duration | Continuous or at short intervals | A few days of review plus report | A fixed test window depending on scope, plus report |
| Result | List of vulnerabilities with severity | Prioritised list of actions per control | Report with demonstrated findings, attack paths and remediation guidance |
| Typical trigger | Day-to-day operations, patch management | First baseline, audit preparation, new person in charge | Customer requirement, insurer, proof of effectiveness, major change |
What does a scan find, and what does it miss?
A scanner sees that a security update is missing on a server. It does not see that the helpdesk group is allowed to reset the passwords of administrator accounts. It does not see that an app registration in Entra ID holds a permission that lets it modify the entire directory, or that a Conditional Access policy contains an exclusion group to which any group owner can add members.
In Microsoft environments, points like these are the most common routes to administrator rights. They are not software bugs with a CVE number; they are configuration and authorisation logic. An assessment finds many of them because it reads the settings. A pentest shows which of them can be chained into a complete path, and therefore which fix actually breaks that path.
Scans also differ from one another. An unauthenticated scan sees a system only from the outside and infers vulnerabilities from version numbers and responses. An authenticated scan, or an agent on the device as with Defender for Endpoint, reads installed software and updates directly and therefore usually gives more accurate results. Ask your IT service provider which type of scan it uses, how often it runs and who evaluates the results. A scan whose results nobody works on only produces a list.
Conversely, it would be wasteful to have a pentester spend paid time on missing updates that a scanner lists in minutes. Each approach has its place.
Which approach does your organisation need?
The following overview helps with the decision:
| Your situation | Recommendation |
|---|---|
| You do not know which systems are unpatched | Introduce vulnerability scanning and feed the results into the patch process |
| Microsoft 365 and Azure have never been reviewed in a structured way | Security assessment |
| Fundamentals such as MFA for administrators or blocking legacy authentication are unclear | Security assessment, then remediation |
| A customer, auditor or insurer asks for a test report | Penetration test with a clearly defined scope |
| You have hardened the environment and want to know whether it holds | Penetration test |
| After a major change (migration, new application, merger) | Assessment of the new parts, then a targeted pentest |
A common mistake is to start with the pentest while fundamentals are still open. The report then lists items that an assessment would have found more cheaply, and the real question about resilience remains unanswered.
How do the three approaches fit together over a year?
For most mid-sized companies we recommend an annual rhythm on three levels:
| Period | Activity | Result |
|---|---|---|
| Continuous | Internal and external vulnerability scanning; monitor the Secure Score in Microsoft Defender and Defender for Cloud | Current patch status, configuration trend |
| First quarter | Security assessment of Microsoft 365, Entra ID, Intune and Azure | Prioritised list of actions |
| Second quarter | Remediation of critical and high items | Documented changes |
| Third quarter | Penetration test, internal and for cloud identity | Evidence of effectiveness, remaining attack paths |
| Fourth quarter | Remediation and retest | Confirmed closure of findings |
The quarters are an example; the order is what matters. Event-driven reviews, for example after a migration, come on top.
This combination matches what regulation asks for. Article 21(2)(f) of the NIS2 Directive and Section 30(2), second sentence, no. 6 of the German BSI Act (BSIG) require policies and procedures to assess the effectiveness of risk-management measures. DORA Article 25(1) lists, among other possible tests, vulnerability assessments and scans, network security assessments, gap analyses and penetration testing. An annual plan that connects and documents these approaches is easier for auditors to follow than a single report.
What do the three levels cost?
Scanning mainly involves licence and operating effort; if you already use Defender for Endpoint Plan 2, the core capability is in place. Wenske CS offers the security assessment, the penetration test for Microsoft 365 and Entra ID, the Active Directory assessment and the Azure assessment at a fixed price. The price is set after the scoping conversation.
Next step
If you do not know where your environment stands, start with the fixed-price Security Assessment. If the fundamentals are in place or a customer asks for a test report, our penetration testing page describes scope and process. For ongoing vulnerability management and monitoring there is Managed Security.
Frequently asked questions
Does the Microsoft Secure Score replace a security check?
No. The Secure Score measures how many of Microsoft’s recommended settings are in place and is useful as a trend value. It does not weight items by your business risk, does not know your operational context and does not assess processes. An assessment puts the values in context and sets priorities.
Is an annual vulnerability scan enough?
Not for patch status. New vulnerabilities are published constantly, and a scan once a year shows only a snapshot. Scan continuously or at short intervals and feed the results into your patch process.
Does an unregulated company need a pentest at all?
Not necessarily every year. It makes sense when customers ask for a test report, when your cyber insurer asks for one, or when you want to know after hardening whether it holds. Without these triggers, an assessment followed by remediation is often the more effective use of the budget.
Can a vulnerability scan disrupt systems?
Modern scanners are generally gentle. Older devices, production equipment or specialised hardware can react sensitively to active scans. Exclude such systems at first, test in a maintenance window and agree the scan with whoever operates them.
Book an introductory call