Skip to content

News

What a penetration test covers, and what it leaves out

A penetration test answers one narrow question well. This article explains what it covers, what it leaves out and what to agree in writing before the work starts.

  • Published on
Abstract pattern of navy and blue shapes on a navy grid: the standard cover of a guide about software, security and IT services.

This article was written automatically on 17 November 2026 from the information on this site and checked by automated rules; it was not reviewed by a person. Tell us if something is wrong

A penetration test, often called a pentest, is a controlled attack on your systems. A security engineer carries it out with your written agreement. Its aim is not to find every flaw. Its aim is to answer one question: can someone with a defined starting point reach something that matters? If you know this, you can buy the right test and read the result correctly.

What a penetration test covers

The scope is the list of things the testers are allowed to examine. Typical subjects are:

  • Web applications: login, user permissions, forms and the interfaces that connect systems (APIs, or application programming interfaces).
  • Networks: servers, firewalls, remote access and the paths between parts of the network.
  • Cloud services: the settings of accounts, storage and access rights.

The starting point matters as much as the scope. A test can start from the outside, like an unknown visitor on the internet. It can start from an ordinary user account, to see what a careless or dishonest user could reach. It can also start from inside your network, to see how far someone could go if one laptop were taken over. Each starting point answers a different question, so choose the one that matches your worry.

The result is a set of findings. Each should name the weakness, show the evidence, describe the impact and recommend a fix.

What it does not cover

A penetration test is useful, but it has limits. Be clear about them before you rely on the result.

  • Anything outside the agreed scope. If a system is not listed, nobody looked at it.
  • Every possible weakness. Time is limited, so testers follow the paths that look most promising. A clean report does not prove that nothing is wrong.
  • The days after the test. The result is a snapshot. A new update, a new account or a new vulnerability can change the picture the next week.
  • People and procedures, unless you ask for them. How your team handles requests, leavers or backups belongs to a wider security assessment.
  • Risky actions. Tests that could disturb your operations, such as overloading a service, are usually excluded by the rules unless you agree otherwise.

A test also does not replace vulnerability management. That is the routine work of finding out which of the vulnerabilities published every week affect the products you run. The test looks at your systems at one moment. Vulnerability management follows them all year.

What to agree before it starts

A good test begins with a written agreement. Check that it states at least the following:

  1. Scope: which applications, networks or services are included, and which are not.
  2. Starting point: outside, ordinary user or inside the network.
  3. Rules: what the testers may and may not do, for example whether they may try to read real data.
  4. Time window: when the work happens, so that your operations can carry on.
  5. Contacts: who on your side can be reached quickly if something unexpected happens, and who on the testers' side.
  6. Handling of results: who receives the report and how it is kept confidential.
  7. Follow-up: whether the testers will check your fixes afterwards.

If your systems are run or hosted by another company, ask whether that company must also give permission. Do this before the test, not after.

How to use the report

A useful report serves two readers. The manager needs a short summary that says what matters and what to decide. The engineer needs detail: evidence, impact and a clear fix for each finding. Ask for both.

Findings should come in order of risk. Start with the ones that would let someone reach something important with little effort. Give each fix an owner and a date. Ask the testers to explain the findings in a conversation, not only in a document. Questions are easier to settle in a short meeting than in a long e-mail thread.

After the fixes are made, check them. A retest is the simplest way to confirm that a problem is really closed and that the fix did not open another one.

Getting help with a test

Pinet's cybersecurity consulting includes penetration testing of web applications, networks and cloud services, alongside security assessments and hardening plans, with a one-page summary and findings in order of risk. You can read more on the page Cybersecurity consulting at Pinet.

More

More news

All news

Have a question about this news?

Write to us. A consultant will answer.