What a penetration test is
A penetration test (PT for short) is a controlled test in which a professional finds weaknesses (vulnerabilities) in an organization’s systems and checks whether they can actually be exploited, the way a real attacker would try. The difference from an attacker is that the test is done with written authorization, within a defined scope and under rules of engagement agreed in advance, and its result is a report that explains what was found and how to fix it.
The goal of a penetration test is not to prove that a “break-in” is possible. The goal is to give the organization a reliable picture of its level of exposure: which vulnerabilities exist, what an attacker could achieve through them, and what needs to be fixed first. A good test ends with a clear understanding of the risk and a work plan, not a long list of problems with no context.
Why an organization needs a penetration test
Most organizations know which systems they installed, but from the inside it is hard to see how everything looks from the outside. A server left open to the internet after an old project, an employee account that was not closed when the person left (offboarding), permissions that were expanded “temporarily” and stayed, a form on the website that does not properly check what is entered into it. Each of these looks small. Sometimes two or three of them together open a path for an attacker all the way to the most sensitive information in the organization.
A penetration test answers a question that automated tools do not answer on their own: what can really be done with these vulnerabilities, in this specific organization. The answer lets you decide where to invest, instead of trying to fix everything at once or fixing what stands out rather than what is dangerous.
There is also a business side. When an incident happens, the organization is asked what it did to test its defenses. The question comes from the Privacy Protection Authority, from the insurance company, and sometimes from large customers already at the contracting stage, in security questionnaires for suppliers (vendor assessment). A documented test, and the remediation carried out after it, are part of the proof that the organization acted responsibly.
Types of tests
A penetration test is defined by what is tested and by how much information the tester receives in advance. By what is tested:
- External test (external PT): everything reachable from the internet, such as websites, VPN connections, mail servers and the firewall.
- Internal test (internal PT): what an attacker can do once already inside the network, for example through an infected employee computer. This usually covers Active Directory, network segmentation (VLANs) and permissions.
- Application test (web application PT): websites, customer portals and APIs, according to the OWASP methodologies.
- Wireless network: the separation between the guest network and the internal network, and how devices connect to the employee network.
- Social engineering: for example a phishing simulation, which checks how employees respond to a suspicious email.
- Cloud configuration review: the security settings of Microsoft 365, Google Workspace or another cloud environment.
By amount of information: in a black box test, the tester starts like an outsider, with no prior information. In a grey box test, the tester receives a regular user account and basic information about the system. In a white box test, the tester gets access to documentation, settings and sometimes code. For most organizations, grey box is the right balance: it saves work on gathering information that a determined attacker would obtain anyway, and puts the effort into the test itself.
How a professional test works
A serious penetration test follows a recognized methodology, so you can know what was tested and what was not. Two common frameworks are PTES (Penetration Testing Execution Standard) and NIST SP 800-115, a guide from the US National Institute of Standards and Technology (NIST) for technical security testing. For web applications, testers work according to OWASP WSTG (Web Security Testing Guide). All of them repeat the same main stages:
- Pre-engagement: defining the goals, the scope and the rules of engagement, and signing the authorization.
- Reconnaissance and mapping: which systems exist, what is exposed and what runs on them.
- Vulnerability analysis: a combination of automated tools and manual testing.
- Controlled verification (exploitation): checking whether the vulnerability can in fact be exploited and what it allows, without disrupting day-to-day operations.
- Reporting: documenting the findings, the impact and the recommendations.
Throughout the process, the tester acts only within the scope approved in writing. Exactly the same actions, when done without authorization, are unlawful intrusion into computer material under the Computers Law. That is why even a severe finding discovered in the middle of the work is reported immediately to the designated contact person, and the tester does not continue “to see where it leads” outside what was agreed.
What you get at the end: the report
The report is the product of the test, and it is worth knowing in advance what should be in it. A good report usually includes:
- Executive summary: a picture of the current state in business language, which lets management understand the level of risk and make decisions.
- Scope and methodology: what was tested, what was not tested and by which method. Without this, you cannot know what the report actually attests to.
- Detailed findings: for each finding, a description, the system where it was found, evidence of its existence and the possible impact.
- Severity rating: usually according to CVSS (Common Vulnerability Scoring System), a scale from 0 to 10 that expresses the severity of the vulnerability, together with the tester’s assessment of the context in the organization.
- Remediation recommendations: what needs to change, at a level of detail that lets the IT team act.
After the report: remediation and retest
A report that goes into a drawer improves nothing. The stage after the report is a remediation plan: a list of findings in order of priority, with an owner for each finding and a closing date. Some fixes are simple, such as updating a version as part of patch management or shutting down an unnecessary service. Others require changes to the infrastructure, such as separating networks, reducing permissions to the minimum required (least privilege) or adding multi-factor authentication (MFA), and they need to be planned so they do not stop the work.
After remediation comes a retest: the tester verifies that each finding was indeed closed and that the fix did not open a new problem. Only then can the finding be marked as handled. When a finding cannot be fixed immediately, for example in an old system the vendor no longer updates (end of life), the decision is documented and a compensating control is added, such as isolating the system in a separate network and restricting access to it.
Penetration testing and regulation
The Privacy Protection Regulations (Data Security) place security obligations on every organization that holds a database of personal information, according to the database’s sensitivity level. For databases at a high security level, they require a risk assessment and periodic penetration tests. Since Amendment 13 to the Privacy Protection Law came into force in August 2025, the Privacy Protection Authority has broader enforcement powers, including imposing administrative fines. The National Cyber Directorate guidance and the ISO 27001 standard also include management of technical vulnerabilities and periodic testing of defenses.
A documented penetration test, together with documentation of the fixes made after it, is practical evidence that the organization actually tested its defenses and acted on what was found. A written procedure alone does not show that. When an incident happens, this is exactly the kind of documentation people ask to see.
Who tests and who fixes
The value of a penetration test is measured by what happens after it. Organizations often receive a detailed report from an outside party, and then discover there is nobody to turn it into work: the findings involve the servers, the firewall, Active Directory and the website, and the people who manage them were not part of the test and do not know the context. The result is partial fixes, or a report that waits for the next test.
That is why the organization’s IT company needs a deep understanding of information security. A partner who can run a test according to a recognized methodology, explain the findings to management, fix them in infrastructure it knows and perform a retest closes the loop. It is just as important that the partner knows the requirements of the Privacy Protection Law, Amendment 13 and the National Cyber Directorate guidance, so that the order of remediation also fits the regulatory exposure. The difference shows on the day an incident happens: an organization that tested and fixed can show what it did, and an organization whose report stayed in a drawer has to explain why it did not act.
Summary
- A penetration test checks whether vulnerabilities can actually be exploited, and is done only with written authorization, within a defined scope and under rules of engagement.
- Define in advance what is tested: external, internal, application, cloud or a combination.
- Require a report with an executive summary, verified and rated findings, and recommendations that can be carried out.
- Build a remediation plan with owners, and finish with a retest.
- Keep all the documentation: it is part of meeting information security requirements.