Two points of view on the same organization
A penetration test examines the organization from an attacker’s point of view. The difference between an external and an internal test is where the attacker starts. In an external test (external PT) the tester is on the internet, like any stranger, and checks what can be seen and obtained from outside. In an internal test (internal PT) the tester starts from inside the network, as if they already have a foothold: an employee’s computer, a network port in a meeting room or a regular user account.
The two tests answer different questions. The external one checks how hard it is to get in. The internal one checks what happens after someone is already in, and how far they can get. An organization that tests only one side gets half the picture, and sometimes the half that matters less to it.
External test: what is checked
The basis of an external test is mapping everything the organization exposes to the internet, known as the attack surface. More than once the list is longer than people thought: a test server that was left open, a management interface of network equipment, or a service opened to a supplier for remote support and forgotten. Among the things an external test checks:
- Exposed services: which services are open to the internet, and whether they are needed.
- Remote access: VPN connections, remote desktop and employee portals, including how authentication works and whether two-factor authentication (MFA) is required.
- Versions and updates: services with known vulnerabilities (CVE) that were not updated as part of patch management.
- Websites and web systems: at a basic level, and if needed in a full application test according to OWASP WSTG.
- Email and domain: settings that make it harder to impersonate the organization’s domain (SPF, DKIM, DMARC).
- The firewall: whether what is actually open matches what the organization thinks is configured.
In an external test it is especially important to confirm that every address and system in scope actually belongs to the organization or is managed for it. A website hosted at a hosting provider or a cloud service may require coordination with the provider, according to its terms, before starting.
Internal test: what is checked
An internal test focuses on what happens inside the office and the network. This is where most of the important assets are: the file servers, the accounting and ERP systems, the backups and Active Directory, which manages all users and permissions. Among the things checked:
- Active Directory: accounts with excess permissions, service accounts with old passwords, and the password policy.
- Separating the network into zones (segmentation, VLAN): whether an employee’s computer can see the servers, the backup and the management interfaces.
- Privilege escalation: whether a regular user can reach administrator rights.
- Lateral movement: whether it is possible to move from computer to computer and from server to server.
- Shared folders (file shares): sensitive data that every employee can access, such as payroll files or customer documents.
- The local administrator password: whether it is the same on every computer, or each computer has its own unique local admin password (LAPS).
- Detection and response: whether endpoint and server protection (EDR) and monitoring detect the tester’s activity.
Why test from the inside too
Some organizations settle for an external test, on the assumption that if it is hard to get in, there is no reason to check what is inside. In practice, quite a few incidents do not start with a break-in through the firewall at all. An employee opens a file from a phishing email, a password leaks from another system, a supplier with remote access is compromised, or a laptop gets infected outside the office and comes back to the network. In all these cases the attacker is already inside, and the external protection plays no part.
That is why an internal test is based on a working assumption called assumed breach: you assume someone is already inside, and check how far they can get. The difference between an organization where one infected computer stays a local incident, and one where a single computer leads to control of the whole network and encryption of every server, lies almost entirely inside the network: in permissions, separation and monitoring.
When you connect the two tests
The greatest value comes from looking at the two tests together. An external finding rated medium, for example a remote access service without two-factor authentication, looks completely different when the internal test shows that a regular user on the network can reach administrator rights. Together they describe a complete attack chain, from the internet to control of the network.
This is also how you set priorities for fixing. The finding with the highest CVSS score is not always the first to handle. Sometimes one link in the chain, such as adding MFA to remote access or reducing the permissions of one service account, breaks the whole path and lowers the risk more than any other fix.
How to decide what fits the organization
The decision depends on how the organization is built and what it exposes:
- An organization with many services open to the internet, such as a customer portal, remote access for employees or hosted servers: an external test is an essential starting point.
- An organization with an office, local servers and Active Directory: an internal test will reveal most of the significant risks, even if the external exposure is small.
- An organization that moved to the cloud (Microsoft 365, Google Workspace): it is worth adding a cloud configuration review, because much of the identity and data lives there.
- An organization that holds sensitive personal data: it is worth covering both sides, because the database itself is usually on the internal network, and the way to it can come from either direction.
In most organizations the practical answer is a combination. You can start with one type according to the most obvious risk, fix, and then move on to the other. The main thing is to define explicitly what was tested and what stayed out of scope, so that a good result in one test is not read as confirmation that the whole organization is protected.
Authorization and scope in both tests
Both types of test are done only with written authorization from an authorized person in the organization, with a defined scope and agreed rules of engagement. In an external test the scope includes an exact list of addresses, domains and systems. In an internal test it includes the network segments, the sensitive systems that require care, and the starting point: a physical connection in the office, a remote connection to a dedicated computer or a regular user account.
An internal test also needs coordination with whoever runs the monitoring (SIEM / SOC): whether the monitoring team and the EDR know about the test in advance, or whether the test also checks if they detect it. Both approaches are legitimate. In both, you need an available contact who can stop the test if something goes wrong, and a way to tell that the activity belongs to the tester and not to a real attacker.
Who fixes the findings
Internal test findings almost always touch the infrastructure itself: Active Directory, network separation, permissions, privileged account management (PAM) and workstation settings. These are exactly the areas the IT company manages day to day. When whoever tests and whoever fixes do not talk to each other, the recommendations stay general and the fixing gets postponed.
An IT partner that understands information security can do both: run the test according to a recognized methodology such as PTES or NIST SP 800-115, and fix the findings in infrastructure it knows, without stopping work in the office. After the fix it runs a retest and confirms the chain is broken. It also knows what is required of the organization under the Privacy Protection Law, Amendment 13 and the National Cyber Directorate guidance, and can document the process so it can be presented when questions are asked.
Summary
- An external test checks how hard it is to get in. An internal test checks how far someone can get from the inside (assumed breach).
- Map everything exposed to the internet before an external test, and confirm that all assets in scope belong to you or are managed for you.
- If there is an office with Active Directory and local servers, an internal test is a practical necessity.
- Read both reports together, and fix first the links that break an attack chain.
- In every test, insist on written authorization, a defined scope and rules of engagement.