Four types of work, four questions
| Work | Question asked | What you get | Limitation |
|---|---|---|---|
| Vulnerability scan | What known weaknesses do my systems have? | A list produced by a tool, often long | False positives, and nothing on what is actually exploitable |
| Security audit | Does my system comply with good practice or with a framework? | Deviations from a reference | Does not prove that a deviation is exploitable |
| External exposure audit | What can be seen of my company from the Internet? | A verified map of what is exposed | Does not attempt to get in |
| Penetration test | How far can a third party go on this target? | Access obtained, with the evidence | A limited scope and a limited time |
The vulnerability scan
An automated tool queries your machines and your applications, and compares what it finds against databases of known vulnerabilities. It is quick, inexpensive and can be repeated effortlessly: that is its great merit.
Its limitations stem from its nature. It produces false positives, which have to be sorted. And it does not understand what your application does: that a user can read someone else’s order by changing a number in the address is something no scan will see.
A scan is not a penetration test, even when the report bears that name. This is the most common confusion, and it is a costly one: you believe you have been tested.
The security audit
An audit compares a system against a reference. Depending on its subject, it covers the configuration of a server, a firewall or a directory service, the architecture of a network, the source code of an application, or the organisation — as is the case for an audit against the ISO 27001 standard.
The auditor works with the information you provide: read access, documentation, interviews. The audit can therefore be thorough on its subject. In return, it says what is badly configured, not what an attacker would do with it.
The external exposure audit
It draws up an inventory of what your company presents to the Internet: domain names and subdomains, including those that have been forgotten, open services, certificates, email configuration, credentials already present in data leaks.
It is carried out without access to your network and without risk to production. It is often the first piece of work to do, because you cannot protect what you do not know exists.
The penetration test
It starts from a specific target and seeks to go all the way: getting in, accessing data, obtaining privileges. It chains weaknesses together as an attacker would, and it reports the evidence of what it obtained.
It is the most demonstrative type of work, and the most dependent on the person carrying it out. It requires preparation: scope, written authorisation, level of information provided.
Which one to choose
- You have never done anything: start with an external exposure audit. It is quick, poses no risk to production, and shows where to focus the effort.
- You are putting an application or a new version online: a penetration test on that application.
- A customer or an insurer asks for evidence: read exactly what is being asked for. "Penetration test" and "scan" are not equivalent, and a scan report presented as a test may be rejected.
- You are preparing for certification or for compliance (ISO 27001, NIS2): an audit against the framework, supplemented by technical tests.
- You want follow-up over time: regular scans, and a test after every significant change.
In what order
- Map External exposure audit
- Fix What is obvious
- Test Targeted penetration test
- Maintain Scans and further tests
- Draw the map: the external exposure audit.
- Fix what is obvious: unnecessary services, overdue updates, open administration access.
- Test what matters: a penetration test on the targets whose compromise would cost the most.
- Maintain: regular scans, and a new test after every significant change.
Having a system tested in depth when its obvious defects have not been fixed amounts to paying a tester to find what an inventory would have shown.
Frequently asked questions
Does a penetration test replace an audit?
No, they complement each other. The audit is more thorough on its subject, since it sees the entire configuration; the test shows what is actually exploitable. A mature organisation does both.
Is a free online scan enough for a small business?
It is better than nothing, and it sometimes reveals an obvious defect. But these tools see only one address at a time, do not look for your forgotten subdomains and do not verify their results.
What about the "red team"?
It is a long exercise that simulates an adversary by every agreed means — technical, human, sometimes physical — to test detection and response as much as protection. It is intended for organisations that have already fixed what audits and tests reveal.
Sources
- ANSSI (the French national cybersecurity agency) — Guide to IT hygiene (in French)
- OWASP — Web Security Testing Guide
This guide provides general information, up to date on the date shown at the top of the page. It replaces neither the official texts nor advice tailored to your situation.

