Side by side
| Vulnerability scan | Penetration test | |
|---|---|---|
| How | Automated tool | Skilled tester, tool-assisted |
| Finds | Known CVEs, missing patches, misconfigurations | Exploitable paths, logic flaws, access control failures, chained attacks |
| False positives | Common | Validated before reporting |
| Frequency | Weekly to monthly, or continuous | At least annually and after major change |
| Output | A long list of findings | A prioritised report with proof, impact and fixes |
| Who runs it | Your IT team, MSP or a managed scanning service | An independent, qualified tester |
| Question it answers | Are we exposed to known weaknesses right now? | What could an attacker actually reach, and how badly would it hurt? |
What a vulnerability scan actually does
A scanner compares what it can see on your systems against a database of known vulnerabilities and insecure settings. It fingerprints software versions, checks for missing patches, flags weak TLS and default configurations, and scores each finding, usually with CVSS. It is fast, cheap to repeat and good at breadth.
Scans come in a few forms, and the differences matter:
- External scans look at your internet-facing hosts and services the way anyone on the internet can.
- Internal and authenticated scans log in to servers and workstations and see far more, including missing operating system and application patches that an unauthenticated scan can only guess at.
- Web application scanners crawl a site and probe for common flaws such as injection and cross-site scripting.
- Cloud posture tools check AWS, Azure or Google Cloud settings against benchmarks, such as public storage, overly broad roles and disabled logging.
The weakness of a scan is judgement. It reports what matches a signature, so you get false positives to weed out, and it can’t tell you whether a medium-rated finding on your customer database matters more than a critical one on a test box. A scan without someone reviewing and acting on the results is not vulnerability management.
What a penetration test actually does
A pen test is a time-boxed engagement in which a tester behaves like an attacker against an agreed scope. A good one follows a recognised methodology, such as PTES, the OWASP Web Security Testing Guide or NIST SP 800-115, and runs in clear phases:
- Scoping and rules of engagement: what’s in and out, test windows, test accounts, who to call if something breaks.
- Reconnaissance: mapping the attack surface, including forgotten subdomains, exposed services and leaked credentials.
- Discovery: scanning and manual analysis to find candidate weaknesses.
- Exploitation and validation: safely proving which weaknesses are real, and how far they lead when chained.
- Reporting: an executive summary, then each finding with evidence, severity, affected assets and a practical fix.
- Retest: confirming the serious findings were actually fixed.
Testers use scanners in the discovery phase, which is why the two are confused. The difference is everything after it: a human deciding what matters, proving it, and showing the business impact.
Types of penetration test
“A pen test” isn’t one thing. Pick the type from what an attacker could reach and what holds your sensitive data.
| Type | The question it answers |
|---|---|
| External infrastructure | What can anyone on the internet find and break into: hosts, services, VPNs, exposed admin panels? |
| Web application | Can a user bypass login, see another customer’s data or abuse a business process? |
| API | Is authorisation enforced on every endpoint, and does the API leak more data than the app shows? |
| Cloud configuration review | Do identity, storage and network settings expose data or let an attacker escalate? |
| Internal network | If a laptop or account is compromised, how far can an attacker move? |
| Mobile application | Does the app store secrets on the device or trust its own back end too much? |
The tester’s starting knowledge matters too. Black-box starts with nothing, like an outside attacker. Grey-box adds user accounts and some context, so the tester can check what a logged-in customer could do; for applications it usually finds more in the same time. White-box adds source code or architecture documents. For how test scope should follow your ISMS scope and risks, see does ISO 27001 require penetration testing?
What scanners miss
Scanners are good at what's known and visible. They struggle with what needs context:
- Broken access control, such as one customer viewing another's data (IDOR)
- Business logic flaws, like skipping a payment step
- Weaknesses that are harmless alone but dangerous chained together
- Authentication and session flaws in custom applications
Broken access control has topped the OWASP Top 10 for years, and it's almost always found by people, not scanners.
Scan, assessment, pen test or red team?
Suppliers use these terms loosely, so read the scope, not the label.
| Service | What you actually get |
|---|---|
| Vulnerability scan | Automated output, often unfiltered |
| Vulnerability assessment | Scans plus human review: false positives removed, findings prioritised, no exploitation |
| Penetration test | Manual testing of a defined scope, with exploitation and proof of impact |
| Red team exercise | A goal-based campaign testing whether your people and monitoring detect and respond to a realistic attack; usually for mature organisations that already pen test |
The common trap is a scan rebadged as a pen test. If the report has hundreds of findings, no evidence of exploitation and no description of what the tester tried by hand, it is a scan.
What compliance frameworks expect, and how often
| Framework | Scanning | Penetration testing |
|---|---|---|
| ISO 27001 | Technical vulnerability management (8.8): you set the frequency in your own procedure, based on risk | Not named, but security testing (8.29) and your risk assessment usually lead to one; common practice is at least annually and after significant change |
| SOC 2 | Vulnerability scanning is a point of focus under the system operations criteria and is routinely tested | Not mandated; listed among the types of evaluation in the monitoring criteria, and routinely requested as evidence |
| PCI DSS 4.0.1 | Internal and external scans at least once every three months and after significant change; external scans by an Approved Scanning Vendor; internal scans authenticated | Internal and external testing at least once every 12 months and after significant change; segmentation testing too, more often for service providers |
| Essential Eight | From Maturity Level One: daily for internet-facing services, weekly for office suites, browsers, email clients, PDF software and security products, fortnightly for operating systems; higher levels extend coverage and tighten patch times | Not required by the maturity model |
Two points get missed. First, under ISO 27001 the frequency you write into your own procedure becomes the audit criterion, so don’t promise quarterly tests you won’t run. Second, the Essential Eight scanning cadence is there to drive patching within set timeframes; scanning without patching on time doesn’t meet it. Our Essential Eight page sets out the scanning each maturity level requires, and the move from the Essential Eight to ASD Essentials covers where the baseline is heading.
For the detail on ISO 27001, read does ISO 27001 require penetration testing? and the Annex A controls behind 8.8 and 8.29. SOC 2 auditors look for the same evidence (see SOC 2 in Australia). Outside formal frameworks, enterprise security questionnaires, cyber insurers and government buyers increasingly ask for the date of your last independent test and whether high-risk findings were fixed.
Which do you need?
| Your situation | What to do |
|---|---|
| Small office, no public applications, mostly SaaS tools | Regular scanning, prompt patching and a configuration review; an external infrastructure test if customers ask for one |
| SaaS company or customer portal | Continuous or frequent scanning, plus an annual grey-box web application and API test and a test after major releases |
| Working towards ISO 27001 or SOC 2 | Scanning in place well before the audit, and a pen test early enough to fix and retest before Stage 2 or the start of the SOC 2 observation period |
| Taking card payments in your own environment | Follow PCI DSS 4.0.1 requirement 11 exactly: quarterly scans and annual internal and external tests |
| Just migrated to the cloud or changed architecture | A cloud configuration review and an external test of the new environment |
What drives pen-test effort
Pen tests are scoped in tester-days, and the effort follows the scope, not the size of your company. The main drivers are:
- The number of applications, APIs, IP addresses and cloud accounts in scope
- The number of user roles and permission levels to test in each application
- How complex the authentication, workflows and business logic are
- Black-box versus grey-box (grey-box takes test accounts and setup, but covers more)
- Whether internal networks, cloud configuration or mobile apps are included
- Whether a retest is part of the engagement
A tight scope built from your risks gives better coverage of what matters than a broad, shallow one. Scanning effort is mostly tooling and someone’s time each week to review and patch. For how testing fits the overall effort of certification, see what drives the cost of ISO 27001.
Common mistakes
- Buying a scan and calling it a pen test, then finding the customer or auditor won’t accept it
- Leaving the main product or API out of scope because it seemed too hard to test
- Testing production without agreed rules of engagement, test windows and a contact for problems
- Filing the report without assigning owners and due dates for each finding
- Skipping the retest, so there’s no proof the critical findings were fixed
- Running scans nobody reads, or ignoring the same medium findings month after month
Getting value from a pen test
- Scope the systems that hold customer data or face the internet
- Choose grey-box for applications, so logged-in risks are tested
- Ask for findings mapped to recognised references such as OWASP and CVSS
- Fix the findings and request a retest, then keep the report as audit evidence
Our penetration testing is black-box or grey-box, follows PTES, OWASP WSTG and NIST SP 800-115, and is written for both your engineers and your auditor. Linking the findings into your risk treatment and audit evidence is the part we take on for clients, alongside our ISO 27001 work.
Next step
Check that scanning is running, someone reads the results and patches land on time. Then decide which systems need a pen test and when, working back from your next audit or customer review. To scope a test, book a free 30-minute consultation.
