← Back to blog

Penetration Testing Methodology: The Phases, Standards and Rules of Engagement

November 20, 2026
Penetration Testing Methodology: The Phases, Standards and Rules of Engagement

A penetration testing methodology is the agreed method a tester follows to attack your systems safely and report what they find. In the UK, the NCSC describes the typical pattern as five stages: initial engagement, scoping, testing, reporting and follow-up. What separates a useful test from an expensive scan is how well the first two stages are done, and whether the report at the end tells you what to fix first. Our CREST-accredited testers at Solusec work directly with you through each of these stages, so you talk to the person doing the testing rather than a salesperson.

This guide is for owners, IT managers and directors at startups, small businesses, schools, colleges and charities who need to buy a test, often against a deadline, and want to know what a sound one looks like.

Solusec
solusec.co.uk
Get Clear, Actionable Test Findings
Solusec’s CREST-accredited testers work directly with you, providing expert-led penetration testing and clear, actionable findings to strengthen your security.
Get a Pentest Quote

Key Takeaways

  • The NCSC's typical pattern is initial engagement, scoping, testing, reporting and follow-up. Follow-up is the last step.
  • Frameworks such as PTES and the OWASP guides give testers a technical checklist. The scope and rules of engagement you agree decide what is tested, when and how.
  • Black, grey and white box describe how much the tester knows beforehand. More knowledge usually means more findings for the time paid for.
  • A good report rates each issue by risk and says how to fix it. It should be usable by a non-specialist.
  • Written authorisation is what makes a test lawful. Never let testing begin without it.

Table of Contents

What a penetration test actually is

A penetration test is a controlled attack on your own systems, carried out by a skilled person with your permission. The GOV.UK Service Manual explains that penetration tests proactively attack systems to find weaknesses and show how easy those weaknesses are to exploit, while vulnerability assessments only identify potential weaknesses.

That difference matters when you are choosing what to buy. A scan lists what might be wrong. A test shows what an attacker could really do with it. We explain the distinction further in our guide to the difference between a penetration test and a vulnerability scan.

The NCSC places testing inside an organisation's wider vulnerability assessment and management process, as a way to judge the likelihood and impact of a weakness being exploited. It is one part of ongoing security work, not a one-off event.

The phases of a penetration test

Searches for "pen test phases" return several lists with different numbers of steps. They describe the same work at different levels of detail. The simplest and most relevant for a UK buyer is the NCSC's five-stage pattern, set out in its penetration testing guidance.

1. Initial engagement

This is the conversation before any work starts. You explain what you are trying to protect and why: a customer contract that demands a test, a product launch, a Cyber Essentials Plus plan, a funder or insurer asking questions. The tester explains what is realistic. Time spent here is what stops you buying the wrong test.

2. Scoping

Scoping fixes which systems are tested and which are not. The NCSC says scoping should involve the people who own the risk, the technical staff who understand the target, and a representative of the testing team. If your IT is outsourced, include that provider early, as they often hold the access details and can approve changes.

For a longer walkthrough aimed at the person who signs the work off, see our guide on how to scope a penetration test.

3. Testing

This is the hands-on stage. Frameworks break it into smaller activities. The Penetration Testing Execution Standard, known as PTES, lists seven sections: pre-engagement interactions, intelligence gathering, threat modelling, vulnerability analysis, exploitation, post-exploitation and reporting. In plain terms the tester:

  1. gathers information about your systems that an attacker could find;
  2. works out which routes in are most likely to succeed;
  3. looks for known and unknown weaknesses along those routes;
  4. tries to exploit them, within the limits you set;
  5. checks what an attacker could reach once inside, such as data or other systems.

That last point is what turns a list of flaws into a business answer: if someone got in here, what would they get?

4. Reporting

The NCSC says a report should identify the security issues, rate the risk each one poses to the organisation or system, recommend fixes, and comment on how accurate and effective your own vulnerability assessment process is. If a sample report only lists tool output with no ratings or fixes, it will not help a small team decide what to do on Monday morning.

5. Follow-up

Follow-up is the last stage in the NCSC's model, and the one most often skipped. It covers explaining findings to your team, answering questions while you fix things, and checking that the fixes work. A report that sits in a folder protects nothing.

Which framework guides a penetration test?

There is no single legal framework that every test must follow. Different standards cover different things, and a competent tester uses several.

FrameworkWhat it is forWho it helps most
NCSC penetration testing guidanceThe UK process for buying and running a test: engagement, scoping, testing, reporting, follow-upAny UK buyer
PTESA seven-section technical standard from pre-engagement to reportingTesters and technical buyers
OWASP guidanceTesting of web applications, mobile apps and APIsAnyone with a website, portal or app
NIST SP 800-115A US technical guide to security testing techniquesTeams wanting technical depth
NCSC CHECK Scheme StandardRequirements for testing government and public-sector systemsPublic-sector suppliers

Where the OWASP testing methodology fits

If you run a website, customer portal or app, OWASP is the framework you will meet most. The OWASP DevSecOps Guideline maps web application testing to the OWASP Web Security Testing Guide, mobile testing to OWASP MASVS and MASTG, and API testing to the OWASP API Security Top 10. It says web testing should consider the OWASP Top 10 (a ranking of the most common web application risks), authentication and authorisation, business logic and API surfaces.

OWASP frameworks mapped to web, mobile and API testing

The NCSC's CHECK Scheme Standard shows how seriously this is taken. For web application engagements it sets minimum elements including API or endpoint enumeration, vulnerability assessment, evaluation against the OWASP Top 10, exploitation and post-exploitation activity.

NIST's guide makes a point that applies to buyers as much as testers: no single technique gives a complete picture, so appropriate techniques should be combined to meet the assessment's needs. This is why automated scanning alone does not answer the questions a test is meant to.

Black, grey and white box: how much the tester knows

The other choice you make is how much information to hand over. The NCSC separates open-box testing, where full information is shared, from closed-box testing, where none is, and also describes scenario-driven tests. OWASP uses three labels:

  • Black box: no prior knowledge. This mimics an outsider with nothing but your public address.
  • Grey box: partial knowledge, such as a user login or architecture documents. This mimics a customer, a supplier or someone with a stolen password.
  • White box: full access to source code, infrastructure or architecture. It is the most thorough for the time paid for.

For most startups and small organisations, grey box is a sensible starting point. The tester spends the budget finding real weaknesses rather than guessing at what your system looks like. Black box suits a specific question: what can a stranger see and reach from the internet?

Choosing the type of test by target

Methodology also changes with what is being tested. The main targets are:

  • External infrastructure: the internet-facing addresses of your firewalls, servers and remote access. Our published price for an external IP test is £3,000 one-off.
  • Websites and web applications: logins, forms, content management systems and APIs. Our published price for a CMS website test is £2,500 one-off.
  • Cloud, mobile and AI systems: each has its own OWASP or vendor-specific checks.
  • Internal networks: what an attacker could do once on a laptop or a guest wifi connection.

Prices for other tests depend on scope, which is why scoping comes before a quote. The details are on our penetration testing page.

Rules of engagement: what the document must say

The rules of engagement are the written agreement on how the test will run. In practice they are part of the scoping stage and the contract. The GDS Way, the UK government's technology guidance, says to agree and record the following:

  • test dates and times;
  • the target areas;
  • exclusions;
  • exploits that are out of scope, such as denial of service (flooding a system until it stops working);
  • the technical scope;
  • the technical documentation and tools required.

The NCSC's CHECK Standard adds that the customer and tester must agree the scope, including the system, threats and threat agents, component technologies and configurations, and types of test. It also says test execution must not deliberately damage the system, must minimise accidental damage, and that the company and customer must stay in contact during execution.

In practice, add these to your own checklist:

  1. Named contacts on both sides, including an out-of-hours number if testing runs outside office hours.
  2. A stop rule: who can pause the test if something slows a live system.
  3. Handling of data: what happens if the tester sees personal data, and how findings are stored and shared.
  4. Third parties: your hosting provider, cloud platform or managed service provider may have their own conditions, so check before testing begins.
  5. A test environment where possible. The GDS Way recommends a separate environment that replicates the live service.

Is penetration testing illegal?

Testing systems you do not own or have not been given permission to test can be a criminal offence in the UK under the Computer Misuse Act 1990. Testing is lawful when the system owner has authorised it. That is why the rules of engagement are signed before any work starts. If a tester needs to go beyond the agreed scope, the safe course is to stop and get the change agreed in writing first. If the systems belong to a hosting company or a customer rather than you, you may not be the person who can authorise testing. If you are unsure, take legal advice before signing.

Is penetration testing illegal?

How often to test

The GOV.UK Service Manual says vulnerability testing should be frequent during service development, not a one-off, and should follow Secure by Design from the discovery phase. NHS England Digital says a test should be done at least annually and include web servers, vulnerability scans and checks that default passwords on network components have been changed.

For most small organisations, the practical pattern is an annual test plus a fresh test after a major change: a new website, a new payment feature, a move to the cloud. We offer annual penetration testing for that rhythm, and continuous penetration testing for teams releasing often.

Building assurance into delivery early, rather than at the end, is also a theme in wider thinking about identity assurance at board level, where oversight of security sits with directors rather than only IT.

What to do when you need a test quickly

Urgent requests usually come from a tender, a customer security questionnaire, an insurer or a funder. The methodology does not change when time is short, but the preparation does. To keep a fast test on track:

  1. Decide the one question the test must answer, such as "can someone reach customer data through our web app?"
  2. Prepare a short scope in advance: addresses, URLs, test accounts and who can approve changes.
  3. Choose grey box so no time is spent on discovery you can supply in minutes.
  4. Have a technical contact available for the days of testing.
  5. Ask for the report format up front, so it satisfies whoever asked for the test.

Because our testers are CREST-accredited professionals and you deal with them directly, scoping questions are answered by the people who will do the work.

Working with Solusec

We are a CREST-accredited provider of penetration testing across web, infrastructure, cloud, mobile and AI, built for startups, small businesses, schools, colleges and charities. You engage directly with the testers and assessors, and our recommendations are written in plain English and sized to what your organisation can act on.

Solusec

If you want a test scoped, start with our penetration testing page. If you are also working towards Cyber Essentials, we are a certification body for it and allied schemes, so you can start with Cyber Essentials and plan your testing alongside it. Our guide to choosing a penetration testing provider sets out what to ask for in any scope.

FAQ

What is the last step of the penetration test methodology?

In the NCSC's model the last stage is follow-up, after reporting. It covers discussing the findings, supporting your fixes and confirming they have worked. Some frameworks, such as PTES, end at reporting, but the follow-up is what turns findings into reduced risk.

Which framework guides penetration testing?

No single framework applies to every test. UK buyers usually start from the NCSC's guidance, while testers draw on PTES, OWASP guides for web, mobile and API work, and NIST SP 800-115 for technical technique. The scope you agree decides which apply.

What should a rules of engagement document include?

It should record the dates, target areas, exclusions, out-of-scope actions such as denial of service, technical scope, named contacts and the tools and documentation needed. It should also say who can stop the test. Both sides should sign it before testing starts.

Do I need black box, grey box or white box testing?

Choose black box to see what an outsider can reach, grey box to model a user or someone with a stolen login, and white box for the deepest review. Grey box is a common starting point for small organisations because it makes the best use of the testing time.

Sources

This article was created with assistance of AI. We never use AI with your data unless given explicit permission.