A penetration testing scope is the written boundary of what testers are authorised to attack: which systems, applications and techniques are in play, and which are strictly off limits. The immediate task for anyone commissioning a test is to produce or approve a signed scope document alongside rules of engagement (ROE), using frameworks like NIST SP 800-115 and OWASP APTS as your baseline, or ask a CREST-accredited supplier such as Solusec to draft one with you.
TL;DR:
- Precise identification of each asset, including IP ranges, domain names, and cloud resource IDs, is critical to avoid scope ambiguity and misinterpretation.
- Restrictions must be clearly documented, specifying excluded systems such as third-party services, databases, and environments like CDEs, with CIDR or resource-based listings, not vague descriptions.
- Well-structured scoping involves defining objectives, stakeholders, targets, impact limits, approvals, and change processes, ideally using standardized worksheets to ensure thoroughness.
- Rules of engagement must explicitly authorize techniques, establish escalation contacts, and include legal and operational safeguards to minimize risks during testing.
- For dynamic cloud environments, listing authorized account IDs and regions is more reliable than relying on IP addresses due to infrastructure volatility.
Table of Contents
- What should a penetration test scope include?
- What is typically excluded from a penetration test?
- How do you scope a pentest step by step?
- Rules of engagement and permission to test
- How does scoping differ by test type?
- What documents does a tester need before starting?
- What are the common scope creep pitfalls?
- Solusec's perspective: scoping is where testing quality is decided
- Getting a properly scoped test from Solusec
- Core standards and guidance referenced
- Sources
- FAQ
What should a penetration test scope include?
A scope document earns its keep only when it names targets precisely enough that a tester cannot misinterpret them. Vague entries like "our website" or "the office network" cause more delayed engagements than any other scoping mistake. Every asset needs an identifier a tester can act on directly.
Group your assets by category first:
- External infrastructure: public IP addresses, IP ranges (CIDR blocks), and fully qualified domain names (FQDNs) for every internet-facing service.
- Cloud resources: account IDs, region boundaries, and resource-level identifiers such as ARNs, rather than IP addresses that shift constantly in autoscaling environments.
- Web and mobile applications: specific URLs, virtual hosts, API endpoints, and app store listings or build identifiers for mobile targets.
- Internal networks: named subnets or VLAN ranges, plus any IoT or operational technology segments you want examined.
- People and premises: named individuals or departments for social engineering, and specific site addresses for physical testing.
Beyond naming assets, the scope must state which techniques are authorised. Will testers work with credentials (a "grey box" approach) or start from zero knowledge? Is active exploitation permitted, or should findings stop at proof of vulnerability? Is social engineering, such as phishing employees, part of the exercise? Each answer changes both the cost and the risk profile of the engagement.
Finally, set temporal boundaries: exact start and end dates, permitted testing windows (business hours versus overnight), and named escalation contacts on both sides who can be reached if something breaks. Cloud-native scoping is particularly brittle when it relies on IP lists alone, because ephemeral resources in dynamic environments make static scoping unreliable within days of signing the document.
What is typically excluded from a penetration test?
Exclusions protect systems that a test could genuinely damage, and they need the same precision as your in-scope list. Ambiguity here is where testers accidentally hit something they shouldn't.
Common exclusions include:
- Third-party services you don't own or control, such as a payment gateway, a hosted email platform, or a partner's API.
- Production databases holding live customer data, unless the engagement is specifically contracted to test them with agreed safeguards.
- The cardholder data environment (CDE) if payment processing is involved, unless it has been explicitly scoped and resourced for that purpose.
- Denial-of-service techniques, which are almost universally excluded from standard penetration tests because they risk taking systems offline.
Write exclusions the same way you write inclusions: by CIDR range, resource ID, or hostname, never by description. "Avoid anything sensitive" is not an exclusion a tester can action; "exclude 10.20.0.0/24 and any host matching *.payments.internal" is.
Pro Tip: If you exclude your cardholder data environment from direct testing, ask for a segmentation check instead. PCI DSS guidance expects segmentation testing to confirm that excluded systems genuinely can't be reached from the tested network, rather than assuming the firewall rules work as intended.
How do you scope a pentest step by step?
Scoping well is a structured conversation, not a form you fill in alone at your desk. Here's the sequence that keeps nothing important off the table.
- Define objectives first. Are you testing for a compliance deadline, after a security incident, or as routine assurance? This shapes everything else, including which standard you scope against, whether that's PCI DSS for payment environments or a general framework for internal assurance.
- Identify every stakeholder. You'll typically need IT operations (who owns the infrastructure), application owners, a legal or compliance representative, whoever owns the budget, and any third-party providers whose systems might be touched indirectly.
- Ask the exact targeting questions. What are the IP ranges or CIDR blocks? What FQDNs resolve to in-scope systems? What cloud account IDs and regions apply? Which techniques are authorised, and which assets are explicitly excluded?
- Establish acceptable impact. Can testers cause service degradation to prove a point, or must every action be non-disruptive? This single question resolves most disagreements before they start.
- Record every approval in writing. A verbal "yes, that's fine to test" from someone in a meeting is worthless if a system goes down and nobody can point to a signature.
- Set a change control process. Scope will shift, a new server goes live, a stakeholder remembers an app they forgot. Agree upfront how changes get proposed, approved and logged, rather than handling it by email thread mid-test.
Use the SANS pen-test scope worksheet as a starting template for the scoping meeting itself. It forces the targeting and exclusion questions into the open rather than leaving them implicit.
Rules of engagement and permission to test
Rules of engagement (ROE) are the document that gives testers legal and operational authority to act, and they sit alongside, not instead of, the scope itself. NIST defines ROE as the pre-test guidelines that set the testing team's authority, responsibilities and operational limits, specifically to reduce the risk of service disruption or unintended damage.
A complete ROE typically covers:
- Who has the legal authority to grant permission (this matters enormously in group structures or when third-party infrastructure is involved).
- Explicit deny-lists of systems, techniques or times that are off limits.
- Named escalation contacts, reachable in real time, for both the client and the testing team.
- Formal legal sign-off confirming the client accepts the risk that testing can occasionally cause instability.
Alongside the ROE sits the permission-to-test document itself, carrying signatures, liability terms, and explicit acknowledgement that even careful testing carries some risk of disruption. Operational safeguards then govern how testing actually runs day to day: rate limiting to avoid overwhelming systems, agreed maintenance windows, a cleanup and rollback plan for anything installed during the test, and clear rules for how evidence (screenshots, extracted data samples) is stored and eventually destroyed. Signed permission and explicit ROE reduce far more than legal exposure. They cut down on disagreements after findings land, which is when scope disputes tend to surface if the paperwork was vague from the start.
How does scoping differ by test type?
The right identifier and level of detail changes depending on what's being tested. Treating every engagement type the same way is one of the more common ways scope documents end up too loose to be useful.

External infrastructure tests work best scoped by public IP address and FQDN, with explicit notes on whether a CDN sits in front of the real infrastructure and whether the upstream provider (hosting company, DNS provider) needs separate notification. An external IP test typically focuses on internet-facing services exactly this way.
Internal network tests need named VLANs or subnets, a clear statement on whether privileged or domain admin accounts are in scope, and whether segmentation between network zones is being validated as part of the exercise.
Web application tests should specify virtual hosts, particular routes or endpoints, and the business flows that matter, checkout, login, admin panels, rather than just a root domain. A CMS website penetration test usually needs this endpoint-level detail to be useful.
Cloud environments are best scoped by resource ID, account boundary and region rather than IP address, since infrastructure can scale up and down within minutes. It's worth explicitly stating which cloud API actions are permitted, for instance whether IAM role assumption is allowed, and forbidding control-plane changes unless specifically authorised.
Social engineering and physical tests need an explicit list of named targets or departments, plus clear data protection limits on what captured information (passwords, personal data) can be recorded and how it's stored afterwards.
Pro Tip: For cloud engagements, always list authorised account IDs and regions by name rather than relying on network ranges. Resource-level identifiers like ARNs prevent testers from accidentally hitting ephemeral infrastructure that spun up after the scope document was signed.
What documents does a tester need before starting?
Good scoping runs faster when the paperwork is ready before the first scoping call, not assembled halfway through it.
- Network diagram showing how in-scope systems connect to each other and to anything excluded.
- Data-flow map for any application handling sensitive data, showing where it enters, moves, and is stored.
- Asset inventory listing every in-scope system with its owner and business criticality.
- Expected services and ports for each host, so testers can flag anything unexpected as a potential finding in itself.
The scope document itself should carry, for every target: its identifier, its owner, its criticality rating, the techniques authorised against it, and any exclusions with a one-line justification for each. Agree deliverables and retest timing upfront too. Most suppliers offer a fix verification retest within a defined window, commonly 30 to 90 days, so build that into your remediation planning rather than discovering it after the report lands.
What are the common scope creep pitfalls?
Scope creep rarely arrives as a dramatic event. It creeps in through small ambiguities that nobody caught during scoping.
- Vague target lists ("the marketing website and related systems") that testers interpret differently from what you intended.
- Mid-test changes requested informally, a new server added by email without updating the signed scope.
- Forgotten third parties, where a vendor's API turns out to be reachable from an in-scope system nobody flagged.
- Unclear CDE boundaries, leaving payment systems either untested where they should be, or tested where they shouldn't be.
- Brittle cloud scoping, where IP-based rules break the moment autoscaling replaces a server overnight.
The fix for all five is procedural, not clever: a signed ROE, explicit deny-lists, pre-action validation before any testing action executes, and a documented scope-change process that everyone actually uses instead of routing around it.
Solusec's perspective: scoping is where testing quality is decided

Most of the value in a penetration test is determined before a single tool runs, during the scoping conversation. Solusec, as a CREST-accredited provider, treats scope validation as a gate rather than a formality: ROE are ingested and checked against target lists before testing begins, and boundaries are enforced throughout the engagement rather than assumed to hold static from day one to the final report.
Where this matters most is in environments that don't stay still, cloud accounts scaling resources hourly, web applications shipping new endpoints weekly. A scope agreed in a checkoff meeting three weeks earlier can already be wrong by the time testing starts, which is why dynamic scope monitoring during the engagement window matters more than a thorough document alone. Fixed-price, annual, and continuous engagement models each demand a different scoping conversation, and getting that conversation right the first time saves far more remediation cost than it appears to at the outset.
— Solusec Admin
Getting a properly scoped test from Solusec
Solusec is the option for organisations that want scoping handled by the people who'll actually run the test, not a sales team reading from a script. If you know roughly what you need tested, an external IP test and a CMS-focused engagement are available, both scoped properly before any testing begins so you're not paying for surprises later.

If your infrastructure changes often enough that a one-off test feels like a snapshot that's already outdated, a continuous penetration testing arrangement keeps scope validation running alongside your actual environment rather than treating it as a once-a-year event. Either way, the starting point is the same conversation: what's in, what's out, and what "done" looks like. Book a scoping call through Solusec's penetration testing page and you'll get a CREST-accredited tester on the call, not an account manager, working through your targets, your exclusions and your timeline before any quote is finalised.
Core standards and guidance referenced
NIST SP 800-115 · OWASP APTS · PCI DSS penetration testing guidance · CREST buyer guide · SANS scope worksheet
Sources
- NIST: Rules of Engagement (glossary)
- Technical guide to information security testing and assessment (NIST SP 800-115)
- PCI DSS: Penetration testing guidance v1.1
- A cyber buyer’s guide to vulnerability assessment and penetration testing (CREST)
FAQ
How do you scope a penetration test?
Start by defining objectives and compliance drivers, then list every target by precise identifier, IP, FQDN, or cloud resource ID, alongside explicit exclusions. Gather sign-off from IT operations, application owners and legal, then formalise everything in a signed scope document and ROE before testing begins.
What are the stages of a penetration test?
A typical engagement follows planning and scoping, reconnaissance, vulnerability identification, exploitation or validation, post-exploitation analysis, reporting, and a follow-up retest once fixes are applied. NIST SP 800-115 frames the early planning and documentation stage as the one that most reduces risk of unintended impact later on.
What does "scope" mean in penetration testing?
Scope is the defined boundary of what a tester is authorised to attack, which systems, applications, networks and techniques are included, and which are explicitly excluded. It's captured in a signed scope document alongside the rules of engagement so both parties agree on the same boundaries in writing.
What are the main steps of a penetration test?
Most engagements move through reconnaissance, scanning, exploitation, and reporting, though the exact breakdown varies by methodology and provider. CREST's guidance notes that penetration testing goes further than vulnerability scanning by actively validating exploitability rather than just flagging weaknesses.
How much does a scoped penetration test cost?
Cost depends entirely on scope size and complexity. Solusec's external IP test costs 3000 £ one-off, and a CMS website test costs 2500 £ one-off, with wider engagements priced after a scoping call establishes exact targets and techniques.
Recommended
- How to Choose a Penetration Testing Provider
- Annual Penetration Testing
- Continuous Penetration Testing
- Fixed-Price Penetration Testing
This article was created with assistance of AI. We never use AI with your data unless given explicit permission.
