External network
Everything reachable from the internet: firewalls, VPNs, remote-access portals, mail servers and exposed services.

Penetration testing is an authorised, simulated attack on your systems that finds the weaknesses a real attacker would use, before one does. Arke Technologies, in Johannesburg, scopes, performs and retests penetration tests for South African organisations, public bodies and schools, and helps your team fix what the test finds.
A pen test (short for penetration test, also written pentest) is a controlled attempt to break into a system with the owner's written permission. Pen testing uses the same techniques as criminals, which is why it is also called ethical hacking, intrusion testing or, informally, penetration hacking; the differences are authorisation, a defined scope and a report.
Penetration testing and ethical hacking sit within the wider field of security testing. IT security testing also includes vulnerability scanning and code review; a penetration test goes further by proving which weaknesses can be exploited. Penetration testing is the closest you get to an attack without suffering one, which is why it anchors the testing side of our cybersecurity services.
Organisations commission a pentest before go-live, after major change, for an audit, tender or customer requirement, or on a regular cycle.

A vulnerability assessment and a penetration test answer different questions, and most organisations need both. Vulnerability and penetration testing are often sold together (VAPT), which works well as long as you know what each part delivers.
| Vulnerability assessment | Penetration test | |
|---|---|---|
| Question | What known weaknesses exist? | What can an attacker actually do with them? |
| Method | Mostly automated scanning, then review | Manual testing, guided by tools |
| Depth | Wide and shallow | Narrow and deep |
| Result | A list of issues rated by severity | Proven attack paths, with evidence |
| How often | Monthly or continuously | Commonly once a year and after major change |
A penetration assessment built on a recent vulnerability assessment is faster and cheaper, because the tester does not spend days rediscovering the obvious. Security vulnerability testing run continuously, with a penetration test once a year, suits most mid-sized organisations. Our cybersecurity assessment and audit covers the vulnerability side.

We scope each test to the systems that matter to you. Common targets:
Everything reachable from the internet: firewalls, VPNs, remote-access portals, mail servers and exposed services.
An AD pentest assumes an attacker has a foothold, such as one phished laptop, and tests how far they can move: Windows pentesting, privilege escalation and paths to domain administrator.
Website penetration testing follows OWASP guidance: authentication, sessions, injection, access control and business logic. Web pentesting covers customer, citizen and learner portals; a WordPress pentest adds the plugins and themes WordPress sites depend on.
Software penetration testing (software pen testing) of applications you build or buy, including REST and GraphQL APIs; a GraphQL pentest checks for over-exposed schemas, missing authorisation and query abuse.
Penetration testing on Azure, AWS and Google Cloud focuses on identity, storage exposure and misconfiguration. Microsoft lets customers test their own Azure resources without asking first, provided they follow its rules of engagement.
A GitHub pentest looks for secrets committed to repositories, weak branch protection and build pipelines an attacker could hijack.
Linux servers (Ubuntu, Debian, AlmaLinux), Microsoft SQL Server databases (an MSSQL pentest), data stores such as Redis that should never answer without authentication, and Kubernetes container platforms.

Wireless networks, guest-network separation and phishing simulations that test staff rather than systems. See email security.
Our method follows OWASP and NIST guidance: the OWASP Web Security Testing Guide for applications (an OWASP pentest, in short) and NIST SP 800-115 for technical testing. The Penetration Testing Execution Standard (PTES) and OSSTMM, the Open Source Security Testing Methodology Manual, are the other widely used references. Following a standard, penetration testing becomes repeatable and comparable from year to year.
01.
Targets, exclusions, testing windows and depth, with your written authorisation before anything starts; testing without it is unlawful access.
02.
What an attacker could learn: domains, exposed services, email formats and leaked credentials.
03.
Automated and manual discovery, then triage to separate real risks from noise.
04.
Controlled attempts within agreed limits, never destroying data or disrupting production without your agreement. Signs of a real compromise move us to incident response.
05.
Findings, evidence, risk ratings and remediation advice for technical staff and management.
06.
We help your team fix the findings, then retest each fix.
For ISO 27001 penetration testing evidence, a retested pen test supports Annex A 8.8 (technical vulnerabilities) and 8.29 (security testing in development and acceptance). See ISO 27001 readiness.
Penetration testing tools are easy to obtain. Penetration testing with Kali Linux, the open-source distribution built for security testing, puts hundreds of them in one place: scanners, web proxies, password crackers and exploitation frameworks. Most are open source, which is why pentesting with Kali, or another Linux distribution, is where most testers learn.
Tools find candidates; people find attack paths. A scanner reports a missing patch; a tester chains it to a weak password and an over-privileged account and shows the route to your finance data. Pentesting and criminal hacking differ in authorisation and intent, not technique, so the value of a test lies in the judgement of the person running it.

A good penetration testing report serves the executive who funds the fixes and the engineer who makes them. Our pentest reporting includes:
Reports are encrypted and shared only with named recipients. The report, the evidence and the retest results are yours to keep and to give to auditors, insurers or a tender panel, including after any agreement with us ends. For clients under a managed agreement, open findings are tracked in the monthly service review until closed, so the audit committee sees progress, in line with King IV Principle 12 and, for public bodies, PFMA and MFMA risk-management duties.
Professional penetration testing is only as good as the pen tester doing it, so ask who will test and what they hold. Certifications for penetration testing are a useful first filter: OffSec's OSCP, CREST's practitioner and registered tester exams, GIAC's GPEN and EC-Council's Licensed Penetration Tester (LPT), which is usually what "licensed penetration tester" means. A certified penetration tester has passed a demanding exam; one with years of varied work has also seen what exams miss. The certifications of the testers on your engagement are listed in our proposal.
A junior (jr) penetration tester is normal in any testing team, provided a senior reviews every finding, so ask who reviews. Some organisations build the skill in-house instead: job boards such as Indeed list penetration tester roles, and Penetration Testing: A Hands-On Introduction to Hacking, by Georgia Weidman, is a common first text. But one in-house penetration tester, ethical hacker or not, rarely covers networks, web, cloud and Active Directory in depth, and is never independent of the team that built the systems.

Penetration testing companies in South Africa range from specialist boutiques to large consultancies. Whichever pen testing companies you shortlist, ask:
A penetration testing service can be bought per test or as a recurring programme; recurring, platform-led testing is often sold as penetration testing as a service (PTaaS). Either way, one contract with us covers scoping, testing, reporting and retest; a virtual CISO can own the programme, and findings that monitoring missed feed into managed detection and response.
Last reviewed: September 2026.
A penetration test is quoted per project once the scope is clear. The price follows the number of systems and addresses, the type of testing (external, internal, web application or cloud), black-box or white-box testing, the number of user roles in an application and whether a retest is included. We send a written scope and quote after a short call; our penetration testing cost guide explains the drivers.

A penetration test should cost what its scope justifies: the number of systems, the type of testing and the depth required. A single website costs far less than an internal network with Active Directory and cloud. Be wary of prices that look too low, which often mean an automated scan with a report attached. We quote per project after scoping.
The five phases are scoping with written authorisation, reconnaissance, scanning and vulnerability analysis, exploitation, and reporting. Scoping sets the rules; reconnaissance gathers what an attacker could learn; scanning finds weaknesses; exploitation proves which ones matter; reporting explains the findings and the fixes. We add a sixth step, the retest, to confirm that each fix works.
No South African law names penetration testing as a general requirement, but several obligations come close. POPIA section 19 requires reasonable technical measures to secure personal information, and testing is how you show they work. PCI DSS requires penetration testing where card data is processed, and auditors, cyber insurers, enterprise customers and tenders often ask for a recent report.
The organisation that owns the system is responsible for commissioning penetration testing and fixing what it finds; the tester is responsible for testing safely, within scope, and reporting accurately. In practice the CIO or IT manager sponsors the test, a system owner signs the authorisation, and the board or audit committee should see the results and remediation progress.
No, AI is not replacing penetration testing, but it is speeding up parts of it. AI-assisted tools help with reconnaissance, triage and report drafting. Chaining weaknesses into a realistic attack, judging what is safe to try in a live environment and explaining the business impact still need an experienced tester. Our rule is that AI recommends and people decide.
Tell us what you want tested and why, and we will send a written scope and quote. Talk to us about a first test or a programme.