A vulnerability scan is automated, broad and repeatable: it checks your whole attack surface against known classes of problems, and you can run it every week. A penetration test is human-led and goal-driven: a tester chains small findings into a real attack, abuses your business logic, and signs a report you can hand to a customer or auditor. Most small websites need continuous scanning first; you need a penetration test when a contract, a standard, or a complex login-driven application demands one.

The two are not competitors. They sit at different points on one spectrum, and the honest answer for most owners is "scan constantly, test manually when someone requires it."

Why this matters for your website

  • What goes wrong: owners buy a one-off penetration test, fix the findings, then run nothing for twelve months while plugins and certificates drift out of date.
  • Or the reverse: a customer questionnaire asks for a penetration test report, and an automated scan PDF gets rejected because it is not an attestation.
  • The fix: decide by need. Use scanning for coverage and cadence, use a pentest for depth and proof, and know which one your buyer or regulator is actually asking for.

What a vulnerability scan actually is

  • What it finds: known, classifiable issues — missing security headers, weak TLS, exposed .git or .env files, outdated CMS and plugin versions matched against the NVD and WPScan, and common OWASP Top 10 flaws such as reflected XSS or SQL injection.
  • What it misses: anything requiring judgement. A scanner does not know that coupon codes should not stack, or that order ID 1042 belongs to another customer.
  • How to use it: run it on a schedule and after every deployment, then treat the output as a work queue, not a verdict.

What a penetration test actually is

  • What it finds: chained and logic flaws — a low-severity information leak plus a weak password reset becoming full account takeover, privilege escalation between roles, or a multi-step checkout bypass.
  • What it produces: scope, methodology, evidence, risk ratings and an attestation letter. Testers usually follow the OWASP Web Security Testing Guide or the Penetration Testing Execution Standard.
  • What it misses: everything that changes after the test ends. A pentest is a snapshot of one week, not continuous coverage.

The spectrum, from free tools to a manual pentest

Five rungs, ordered by depth and cost. Read each as: what it finds | what it misses | cost order | frequency | who needs it.

  • Free single-purpose tools — one signal, such as headers or SPF/DMARC | everything outside that signal | free | any time | anyone doing a first sanity check.
  • Passive external scan — DNS, mail authentication, TLS, headers, cookie flags, exposed files, CMS version | anything needing an active request | very low | weekly or monthly | brochure sites, agency triage.
  • Active automated scan — injection, XSS, misconfiguration, vulnerable plugins with CVE correlation, exposed secrets and source maps | business logic, chained auth bypasses | low to moderate, usually a subscription | monthly and after each release | stores, SaaS, any site with forms or logins.
  • AI-assisted assessment — authenticated areas, multi-step flows, IDOR chains, plus human validation of findings | novel research, social engineering, formal attestation | moderate per assessment | quarterly or per major release | SaaS with roles, payments or customer data.
  • Manual penetration test — creative chained exploits, logic abuse, bespoke attack paths, signed report | coverage between engagements | highest | annually or on major architecture change | compliance scope, enterprise buyers, regulated data.

Which do you need? A guide by situation

  • Brochure or marketing site: passive scanning on a schedule is enough. Your real risks are an outdated CMS, an expired certificate and a hijackable subdomain.
  • WooCommerce store: active automated scanning, monthly and after plugin updates. Payment data usually lives with your processor, but coupon, cart and checkout logic is yours to test.
  • SaaS with logins: authenticated scanning plus periodic deeper assessment. Access-control flaws between user roles are the defining risk and rarely show up unauthenticated.
  • Agency with client sites: broad passive scanning across the portfolio, with active scans on the sites that take payments or store personal data.
  • Compliance or a customer questionnaire: read the wording. If it says "penetration test" performed by a qualified tester, commission one — an automated report will not satisfy it.

When scanning does not replace a human test

Some standards require both, separately. PCI DSS, for example, mandates regular internal and external vulnerability scanning and penetration testing as distinct requirements; check the current text at the PCI Security Standards Council. Enterprise procurement and cyber insurance forms often ask for a dated attestation letter, which only a human engagement produces.

If nobody is asking and you handle modest amounts of data, continuous scanning plus fast patching buys you more real-world safety per unit of budget.

How to check your own site

Start free and unauthenticated. Run the Security Headers Checker, the SPF & DMARC Checker and the Website Security Check — no account needed. For a fuller walkthrough, see our guide on how to check your website security.

Scantinel (scantinel.ai) is an automated website security assessment platform that runs OSINT, network, and OWASP Top 10 web vulnerability tests against ownership-verified assets and delivers actionable, business-owner-legible reports. On the spectrum above, the three scans map like this:

  • The Essential scan sits on the passive rung: DNS, SPF/DKIM/DMARC, TLS, security headers, cookie flags, exposed files, CDN/WAF detection and CMS version checks, in minutes.
  • The Full scan is the active rung: subdomain and origin-IP discovery, port and service fingerprinting with CVE correlation, a headless crawl, an active OWASP Top 10 scan, CMS deep scans and supply-chain checks.
  • The Premium AI scan adds agentic, non-destructive probing for business-logic flaws and IDOR chains, authenticated scanning with test credentials, human analyst validation and one retest.

Being direct: the Premium AI scan is a deeper assessment, not a from-scratch penetration test, and it does not produce a tester's attestation. See how the scan works or see Scantinel's scans, and read the broader website security guide for context.

Key takeaways

  • Vulnerability scanning gives breadth and cadence; penetration testing gives depth and proof.
  • Scanners find known issue classes; humans find chained and business-logic flaws.
  • Match the rung to the risk: passive for brochure sites, active for stores, authenticated for SaaS.
  • If a contract or standard names a penetration test, no automated report substitutes for it.
  • A yearly pentest without ongoing scanning leaves eleven months uncovered.

Frequently asked questions

Is a vulnerability scan the same as a penetration test?

No. A vulnerability scan is automated and matches your site against known issue patterns; a penetration test is a human engagement with defined goals, manual exploitation and a signed report. Scans tell you what is probably wrong, pentests demonstrate what an attacker can actually do.

How often should I run a vulnerability scan?

Monthly is a sensible baseline for most sites, and weekly if you run WordPress with many plugins. Also scan after every deployment, plugin update or DNS change, because those are the moments new exposure appears.