Why scope decides the value of a penetration test
A penetration test only finds what it is pointed at. If the scope lists a marketing website but your real exposure sits in a customer API, an admin portal or a cloud storage bucket, the report can come back clean while the actual risk is untouched. Scope is the single biggest factor in whether a test is worth the money.
A good scope starts from the business, not from an IP range. Ask what would hurt most if it were taken over, leaked or taken offline, then work back to the systems that make that possible.
Step 1: list what the business depends on
Before choosing test types, write down the assets that matter. Include the ones people forget:
- Customer-facing web applications and mobile apps, including their backend APIs.
- Identity systems: single sign-on, admin consoles, service accounts and API keys.
- Third-party integrations such as payment providers, CRMs and messaging services.
- Cloud accounts and the services running in them: storage, databases, containers and serverless functions.
- Internal networks, remote access and any systems that hold regulated or customer data.
Step 2: choose the attacker perspective
An external test assumes an attacker with no access, starting from the internet. An internal or assumed-breach test starts with a foothold, such as a compromised laptop or a low-privilege account, and asks how far an attacker could go. Most organisations need both, but if budget is limited, start with the perspective that matches your most likely threat. Internet-facing SaaS products usually start external and authenticated, while organisations with large offices and legacy networks often learn more from an internal test.
Step 3: decide the depth for each target
- Black box: no information provided. It is closest to a real outsider but spends time on discovery.
- Grey box: test accounts and some documentation. This is usually the best value for web applications and APIs, because testers go straight to business logic and access control flaws.
- White box: source code or architecture access. It gives the deepest coverage and suits high-risk or regulated systems.
Step 4: include APIs, cloud and mobile when they exist
Modern products rarely live in one web page. If your mobile app talks to an API, test the API directly, because that is where authorisation bugs appear. If your product depends on managed cloud services, include configuration review of IAM roles, storage permissions and exposed services, since many serious findings are misconfigurations rather than code flaws.
Step 5: agree the rules of engagement
Put these in writing before testing starts, so the work is safe and the report is useful:
- Testing window, including any times to avoid such as month-end processing.
- Out-of-scope systems and third-party services you cannot authorise testing against.
- How far testers may go to prove impact, for example whether they may read sample data or only demonstrate access.
- Who to contact immediately if a critical issue is found mid-test.
- Whether social engineering or physical testing is allowed.
- Retest arrangements, so fixes can be verified afterwards.
A short scoping checklist
- Have we listed every application, API and cloud account that holds customer data?
- Do we know whether we need an external view, an internal view, or both?
- Will testers get accounts for each user role, so they can test access control between them?
- Are third-party systems either excluded or authorised in writing?
- Have we agreed the proof-of-impact rules and an emergency contact?
- Is a retest included to confirm remediation?
Common scoping mistakes
- Testing only the public website and leaving APIs and admin portals out.
- Scoping by IP count instead of by what the business would lose.
- Giving no test accounts, so testers cannot see anything behind the login.
- Treating the report as the finish line instead of planning remediation and a retest.