What does a penetration test cost? Price factors, scope and fixed prices
Key takeaways
- The price of a penetration test follows from its scope: the number of systems, applications and identities, the test type, internal or external perspective, reporting depth and retesting. The day rate on its own says little.
- Offers only become comparable once objectives, method, person-days, report content and retest are stated in writing.
- An automated vulnerability scan with a tool export is not a penetration test, even when it is sold as one.
- At Wenske CS, the fixed price is confirmed in writing after the scoping conversation and before testing begins.
The question of price comes up at the start of almost every introductory call about a penetration test, and it is a fair one. There is usually a specific trigger: a major customer asks for a current test report, the cyber insurer wants to know when the last pentest took place, or NIS2 and DORA require organisations to check that their security measures are effective. Management needs to be able to justify the spend; IT needs a result it can work with. This article explains, from a penetration tester’s point of view, what drives the price, how to compare offers and when a less expensive security assessment is the better first step.
What is a penetration test, and where does the effort go?
A penetration test (pentest) is an authorised, simulated attack. Testers do not stop at finding weaknesses; they try to exploit them and chain them together until it becomes clear how far a real attacker would get. The result is evidence: this path leads from an ordinary user account to domain administration, this misconfiguration opens access to confidential data.
Tools provide raw data. In Active Directory, BloodHound shows possible attack paths; for Microsoft Entra ID, testers use tools such as ROADtools or AzureHound. Whether a path can actually be walked, what conditions it requires and how to demonstrate it without causing damage is manual work. That is where the effort lies, which is why pentests are estimated in person-days. At its core, the price is the number of person-days a given scope requires.
Which factors determine the price?
Scope: systems, applications and identities
The largest cost driver is the scope, meaning the boundary of what is tested. For an external test, the public IP addresses and services count; for web applications, the number of functions and user roles; for Active Directory, the number of domains and trusts. In Microsoft 365 and Entra ID, effort depends less on the number of mailboxes than on configuration complexity: how many Conditional Access policies exist, how many app registrations hold far-reaching permissions, and whether identity is hybrid through Entra Connect. Azure adds subscriptions, role assignments and internet-facing services.
Test type: black box, grey box, white box
| Test type | What the tester receives up front | Effort | Typical use |
|---|---|---|---|
| Black box | Only target and boundaries | High, much time spent on reconnaissance | Simulating an attacker with no prior knowledge |
| Grey box | Test accounts, documentation, contacts | Medium | Standard choice for organisations that want to find weaknesses efficiently |
| White box | Full configurations, source code where relevant | Predictable, very deep review | Critical applications, architecture review |
For most mid-sized companies we recommend grey box. A black-box test spends paid testing time discovering information your IT team could have provided in an hour, and that time is then missing from the actual testing.
External or internal perspective
An external test examines what is reachable from the internet: sign-in portals, VPN gateways, published applications. An internal test starts from an already compromised workstation or user account (assumed breach) and asks how far an attacker can get from there. In Microsoft environments the internal view is often more revealing, because attackers today usually get in through stolen credentials or phishing rather than an open firewall port. Both perspectives can be combined; each additional perspective adds person-days.
Report, retest and timing
A good report contains a management summary and, for each finding, a risk rating, reproducible evidence and remediation guidance. This takes time, but it is the part that auditors, customers and insurers will read later. A retest confirms after remediation that the gaps are closed; some providers include it, others charge separately. Testing outside business hours, short-notice dates or coordination with several service providers also add effort.
How do you compare offers?
Two offers at different prices are often simply two different services. Ask every provider the same questions:
- Which systems, applications and accounts are in scope, and what is explicitly excluded?
- Which test type and which perspective (external, internal, cloud identity)?
- How many person-days are planned, and who will do the testing? Does the provider use subcontractors?
- Which methodology is followed, for example the OWASP Web Security Testing Guide for web applications?
- What does the report look like? Ask for an anonymised sample report.
- Is a retest included, and within what period?
- How are critical findings reported during the test, and what abort criteria apply?
- Where are test data and findings stored, and when are they deleted?
A fixed price after a scoping conversation gives you budget certainty. It requires the scope to be described properly in advance. Time-and-materials pricing makes sense when the scope only becomes clear during the test, but it then needs a cap.
How do you recognise a scan sold as a pentest?
Automated vulnerability scanners have their place. They check many systems quickly for known vulnerabilities and missing updates. They do not, however, chain findings into attack paths, and they understand neither authorisation logic nor business processes. A scan sold as a pentest usually shows these signs:
- The price is fixed without a scoping conversation, often with “unlimited targets”.
- The report is a tool export with hundreds of CVE numbers, without context and without evidence that a finding is exploitable.
- There is no statement about attack paths, for example from user to administrator.
- Nobody asks for written authorisation, test accounts or an emergency contact.
- The testers are not named.
The risk lies less in the lost budget than in a false sense of security. Auditors and demanding customers generally recognise a scan report and will not accept it as evidence of a pentest.
When is a security assessment enough?
A security assessment compares your configuration with recognised baselines, such as the Microsoft cloud security benchmark or the BSI IT-Grundschutz baseline, and delivers a prioritised list of actions. It does not actively attack. It is the right starting point if no structured review has taken place yet or if fundamentals are still open: administrators without MFA, legacy authentication still allowed, permanent Global Administrators. We see these points frequently in assessments. A pentest would find them too, only at a higher price.
A pentest makes sense once the fundamentals are in place and you want to show that they hold against a real attack, when a customer asks for a test report, or after major changes to the environment. Our recommendation for most mid-sized companies: assessment first, then remediation, then the pentest.
On the regulatory side, Article 21(2)(f) of the NIS2 Directive, and in Germany Section 30(2), second sentence, no. 6 of the BSI Act (BSIG), require policies and procedures to assess the effectiveness of risk-management measures. Neither prescribes a pentest by name, but a pentest is one of the most direct forms of evidence. DORA is more specific: financial entities other than microenterprises must, under Article 24(6), ensure at least yearly that appropriate tests are conducted on all ICT systems supporting critical or important functions, and Article 25 lists penetration testing explicitly as one of the test types.
What does a penetration test cost at Wenske CS?
We work with fixed prices that are confirmed in writing after the scoping conversation and before testing begins. The price depends on the size and complexity of your environment.
The result includes a management summary, prioritised findings with evidence and a mapping to NIS2, ISO 27001 and BSI IT-Grundschutz. The retest is optional.
Next step
If you need a test report for customers, auditors or insurers, our penetration testing for Microsoft 365, Entra ID and Azure page describes scope and process. If the fundamentals have not yet been reviewed, the fixed-price security assessment is the less expensive starting point. In the introductory call we clarify the trigger and scope and tell you which route fits.
Frequently asked questions
How often should a penetration test take place?
An annual cycle is common, plus a test after major changes, such as a migration to Microsoft 365 or a new customer-facing application. For financial entities under DORA other than microenterprises, yearly testing of systems that support critical or important functions is mandatory; the appropriate test type follows from the testing programme.
Is the retest included in the price?
At Wenske CS the retest is optional and agreed separately. Ask about it specifically with every offer, because without a retest there is no evidence that the findings have actually been fixed.
Can our IT service provider carry out the pentest itself?
It can, but the provider would then be testing its own work. An independent tester does not have this conflict of interest. DORA Article 24(4) explicitly requires tests to be undertaken by independent parties, whether internal or external.
Will a pentest disrupt day-to-day operations?
A properly coordinated pentest is planned so that operations continue. Tests that could affect availability only take place with explicit approval; load tests and denial-of-service scenarios are excluded by default. Test window, emergency contact and abort criteria are agreed in writing before testing starts.
Book an introductory call