Last updated: November 18, 2026
Anyone can call themselves a penetration tester. A six-week course, a Kali Linux laptop and a confident LinkedIn profile are enough to start selling “pentests” to companies that do not know what to look for. On the other side of the market are testers who have spent decades writing exploits, reading packet captures and breaking systems their builders swore were safe. Both send proposals that look the same on paper.
The difference shows up later. A weak test produces a rebranded scanner report and a false sense of security. A strong one finds the chained, business-logic flaws that attackers actually use. This guide walks through how to tell them apart before you sign, what to ask, and which red flags should end the conversation.
Penetration testing is an unregulated trade in most countries. There is no license to practice, so the label covers everything from fully manual adversary simulation to an automated scan with a cover page. Buyers usually compare proposals on price and turnaround, which rewards the vendors doing the least work.
The result is a market where two quotes for “an external and web application penetration test” can differ by a factor of ten and still describe completely different services. If you are budgeting, our breakdown of penetration testing cost explains what drives those numbers. The rest of this guide is about making sure you are buying the real thing.
| What To Check | Good Sign | Red Flag |
|---|---|---|
| Who does the testing | Named testers with verifiable experience | “Our team” with no names or bios |
| Certifications | Hands-on exams such as OSCP, OSEP, OSWE, CREST CRT/CCT, GPEN | Only multiple-choice or vendor-awarded badges |
| Methodology | Documented approach mapped to PTES, OWASP WSTG or NIST SP 800-115 | “Proprietary process” they will not describe |
| Scoping | Detailed questions before any price is given | Flat price per IP with no discovery call |
| Sample report | Redacted real report with reproduction steps | Refuses to share one, or shares a scanner export |
| Manual vs automated | Clear split, with most hours spent on manual testing | Vague answers about how much is tool output |
| Retesting | Retest of fixed findings included or clearly priced | Retest not mentioned |
| Insurance and legal | Professional liability cover, signed rules of engagement | Starts testing without written authorization |
This is the single most important question, and the one most buyers forget to ask. Many firms sell with senior people and deliver with juniors, or quietly subcontract the work. Ask for the names of the testers assigned to your engagement, their years of hands-on offensive experience, and whether they are employees or contractors.
Good firms answer this without hesitation. Strong testers often have a public trail: CVEs they have disclosed, conference talks, open-source tools, CTF results or bug bounty rankings. You do not need all of these, but you should be able to verify that the person touching your environment has done this before.
Certifications are not everything, but they tell you how someone was tested. Practical, hands-on exams that require compromising live machines carry much more weight than multiple-choice tests. Offensive Security’s OSCP, and its advanced siblings OSEP, OSWE and OSED, are widely respected for this reason, as are CREST’s CRT and CCT qualifications, which are the standard in the UK, and GIAC’s GPEN and GXPN.
Be cautious when a firm leans heavily on entry-level or multiple-choice credentials as proof of seniority. For a deeper comparison, see OSCP vs PNPT vs CEH and our OSCP certification guide. At the company level, CREST accreditation and similar schemes show that a firm’s processes have been audited, not just its people.
Scanners find known vulnerabilities quickly and cheaply. They do not find broken access control between two user roles, a payment flow that can be replayed, or three low-severity issues that chain into full account takeover. Those findings come from a human who understands how your application is supposed to work and then makes it do something else.
Ask what percentage of the engagement hours are manual testing and which tools they run. A credible firm will tell you plainly. If the answer is evasive, you are probably buying a scan. Our comparisons of automated vs manual penetration testing and vulnerability scanning vs penetration testing cover where each one fits.
A real penetration test is priced from the scope: how many applications, user roles, API endpoints, IP ranges, cloud accounts and testing days. A firm that sends a fixed price before asking any of those questions is either padding the quote or planning to run a tool and stop.
Expect a discovery call, a scoping questionnaire and a written statement of work that names the targets, the testing window, the test type (black, grey or white box) and what is out of bounds. That document protects you as much as it protects them.
The report is what you are paying for. Ask for a redacted sample from a real engagement. A strong report has an executive summary a board member can read, followed by technical findings that each include severity, affected assets, clear reproduction steps, evidence such as screenshots or request and response pairs, business impact and specific remediation advice.
Watch for reports that are mostly pasted scanner output, findings with no proof of exploitation, or generic advice such as “apply patches.” Our guide to the penetration testing report shows what a complete one should contain.
Ask how they approach your specific stack. A firm testing a SaaS product should talk about multi-tenant isolation, authorization between roles and API abuse. One testing a cloud environment should talk about IAM misconfigurations and privilege escalation paths, not just open ports. If you are adopting AI features, ask how they test for prompt injection and data leakage through LLM integrations.
Specialists exist for most of these areas. If your needs are narrow, our lists of cloud penetration testing companies, API penetration testing companies, mobile app penetration testing companies and AI red teaming companies are a good place to start a shortlist.
If the test is driven by SOC 2, PCI DSS, HIPAA or ISO 27001, tell the vendor up front. Auditors have specific expectations about scope, frequency and evidence, and a report that misses them can mean paying for a second test. Firms that regularly support audits will know what your auditor wants to see. See penetration testing for compliance for the details, or start with our lists of SOC 2 penetration testing companies and PCI penetration testing companies.
Finding issues is half the job. Ask whether a debrief call is included, whether testers will answer your developers’ questions during remediation, and whether a retest of fixed findings is part of the price. A retest letter confirming that critical and high findings are closed is often exactly what customers and auditors ask for.
Also ask how often they recommend testing. The honest answer depends on how fast your environment changes. Our guide on how often you should run a penetration test covers the usual triggers.
Before any testing starts, you should have a signed authorization and rules of engagement that define targets, timing, emergency contacts and how the tester will handle sensitive data they find. Ask whether the firm carries professional liability and cyber insurance, how they store and destroy your data after the engagement, and what happens if they accidentally take a production system down.
A firm that wants to start without paperwork is a firm you do not want inside your network.
This is the single most important question, and the one most buyers forget to ask. Many firms sell with senior people and deliver with juniors, or quietly subcontract the work. Ask for the names of the testers assigned to your engagement, their years of hands-on offensive experience, and whether they are employees or contractors.
Certifications are not everything, but they tell you how someone was tested. Practical, hands-on exams that require compromising live machines carry much more weight than multiple-choice tests. Offensive Security’s OSCP, and its advanced siblings OSEP, OSWE and OSED, are widely respected for this reason, as are CREST’s CRT and CCT qualifications, which are the standard in the UK, and GIAC’s GPEN and GXPN.
Scanners find known vulnerabilities quickly and cheaply. They do not find broken access control between two user roles, a payment flow that can be replayed, or three low-severity issues that chain into full account takeover. Those findings come from a human who understands how your application is supposed to work and then makes it do something else.
A real penetration test is priced from the scope: how many applications, user roles, API endpoints, IP ranges, cloud accounts and testing days. A firm that sends a fixed price before asking any of those questions is either padding the quote or planning to run a tool and stop.
This guide was last updated on November 18, 2026. We revisit it as policies, prices and programs change.
The gap between a seasoned tester and someone fresh out of a short course rarely shows in a sales deck. It shows in the scoping call, the sample report and the willingness to name the people doing the work. Ask the questions above, insist on evidence, and treat a cheap, fast, unscoped quote as the warning it is.
When you are ready to build a shortlist, start with our list of top penetration testing companies, or narrow it down by region with our picks for the UK and New York.