Key takeaways
- Define scope in terms of systems and protocols, not just network zones, and write exclusions with the same precision as inclusions.
- Every OT/ICS pentest must have a written safety plan that names prohibited actions, halt conditions and the person authorised to stop the engagement.
- A credible proposal separates passive from active testing, references the MITRE ATT&CK for ICS matrix and delivers OT-specific remediation guidance.
- Operations staff must sign off on the safety plan alongside the IT and security teams.
- A vendor who cannot name the industrial protocols they test against, or who has no OT sample report, is not qualified for this engagement.
Why buying an OT/ICS pentest differs from buying an IT test
An IT penetration test and an OT/ICS pentest share a methodology at their core, but the procurement process for the two is not the same. The systems being tested in an OT engagement control physical processes: a valve, a conveyor, a power relay, a pump. A misconfigured test against a corporate web server may cause a service outage lasting minutes. The same error against a PLC managing a cooling circuit carries consequences that an IT team cannot resolve by restarting a service.
The tester's qualifications, the scope definition, the safety constraints and the report format all need to be evaluated differently. A vendor who performs well on infrastructure penetration testing may have no exposure to industrial protocols such as Modbus or DNP3, and that gap is not always visible in a generic proposal. This guide covers what to ask for, what to exclude from scope and how to read a proposal for genuine OT competence.
The procurement difference is not bureaucratic. In an OT environment, an error is measured in operational disruption, not service recovery time. Understand that before you choose a vendor.
Preparation
Scope definition
Systems, protocols and exclusions agreed in writing
Safety planning
Prohibited actions, halt conditions, authorisation chain
Authorisation
Signed by IT, operations and the testing firm
Testing
Passive recon
Traffic capture and protocol analysis only
Active testing
Only if approved and within the safety plan
Reporting
Findings report
OT-specific remediation, not just CVSS scores
Retest
Verified fixes confirmed in writing
Defining scope: what to test and what to protect
OT environments contain several distinct layers: field devices such as PLCs and RTUs, the control layer of HMIs and engineering workstations, the supervisory layer of SCADA systems and historians, and the connections between those layers and the corporate IT network. A single engagement rarely covers all of them, and trying to do so increases both cost and operational risk without proportionate return. Start by identifying where you want assurance, not by listing everything that exists.
The IT-OT boundary is almost always worth testing. The path from a domain-compromised workstation into the control network, the kind of chain described in engineering workstation to PLC: how control logic tampering happens, can often be assessed through passive traffic analysis and credential testing without interacting with live control devices. That scope stays within safe operating bounds and produces findings that operations staff can act on.
Define exclusions with the same precision as inclusions. Sending write commands to PLCs, modifying controller setpoints and testing during production hours should be explicitly excluded unless you have isolated test hardware or a confirmed maintenance window. A vendor who treats these exclusions as negotiating positions is not a vendor to shortlist. The exclusions protect your operation and are not obstacles to a thorough assessment.
Scope creep is a risk in OT engagements in both directions. A scope that expands during the engagement to include live device testing not agreed in advance creates operational risk. A scope that shrinks because the vendor is uncomfortable with the OT environment produces findings that do not reflect the real attack surface. Agree the scope in writing, have it signed by both sides and hold to it.
Safety protocols the engagement must follow
Every OT/ICS pentest must have a written safety plan before any testing begins. The plan names the systems in scope, the actions that are explicitly prohibited, the conditions under which testing will be paused or stopped, and the person authorised to call a halt. This document is part of the authorisation, not an appendix to be reviewed after the engagement starts.
Passive reconnaissance, capturing and analysing network traffic without sending any commands, carries far lower operational risk than active interaction. It can reveal unencrypted Modbus sessions, unauthenticated HMI connections and protocol anomalies without putting any device at risk. Active testing against PLCs or RTUs, if it is required at all, should be conducted against isolated or decommissioned hardware, or reserved for a maintenance window with operations staff present.
Operations staff, the engineers who run the control system daily, must be involved in safety planning and must sign off on the plan before work begins. They will know which devices are running near capacity, which maintenance windows are genuinely available and which constraints the IT team may not know about. Their sign-off is part of the authorisation document, not a courtesy notification.
What a credible OT/ICS pentest proposal must contain
A proposal that lists tools, hours and a standard CVSS finding format is not sufficient for an OT/ICS engagement. The document must show that the vendor understands the technical requirements and the operational constraints of the environment. The table below shows the elements a credible proposal must contain, what to look for in each and what signals a firm is not equipped for this work.
The methodology section should reference the MITRE ATT&CK for ICS matrix and NIST SP 800-115 or equivalent technical testing guidance. Those references tell you the vendor maps findings to a recognised framework, so you can track the same attack categories across engagements and compare results year on year. A vendor who produces a bespoke finding taxonomy with no external anchor makes that comparison impossible.
The deliverable section is where proposals most often disappoint. Ask whether remediation guidance will be written for an OT context: a high-severity finding against a 15-year-old PLC that cannot be patched requires a network segmentation or monitoring recommendation, not a patch deadline that cannot be met. Reviewing what your OT monitoring should catch during an ICS pentest alongside the proposal gives you a baseline for the detection coverage a good engagement should address.
| Proposal element | What to look for | Red flag |
|---|---|---|
| Scope statement | Lists systems by type (HMI, PLC, historian), protocol and network zone | Named only as 'OT environment' with no further detail |
| Testing boundary | Separates passive (read-only) from active (write or command) testing with rationale | Does not distinguish between the two |
| Safety plan | Names prohibited actions, halt conditions and the authorisation chain by role | Absent or deferred to a separate document after engagement start |
| Methodology reference | Cites MITRE ATT&CK for ICS or NIST SP 800-82 alongside general testing frameworks | No framework reference at all |
| Report format | Includes OT-specific remediation, not just CVSS scores and patch deadlines | Standard IT report template with OT terminology substituted |
| Retest provision | Explicitly offered for critical and high findings at no extra cost | Absent or quoted as a separate engagement |
Evaluating whether a vendor is qualified for this work
OT security is a narrow specialism. Ask the vendor directly to name the industrial protocols they test against: Modbus, DNP3, EtherNet/IP, Profinet, IEC 61850 and BACnet are the common ones across industrial and building management contexts. A firm with a red team background and no OT work is not qualified for this engagement regardless of their IT record. Purple teaming may help assess detection coverage in a later exercise, but the initial pentest requires domain knowledge that general IT methodology does not provide.
Request a sanitised sample report from a comparable OT engagement before the shortlisting stage. A strong OT report contains network topology diagrams of the control network, protocol-level findings with traffic captures where relevant, and remediation advice that distinguishes patchable vulnerabilities from those requiring compensating controls. A report that reads as a standard web application assessment with OT terminology inserted shows the firm has not adapted its practice to industrial environments.
References to the NIST Cybersecurity Framework in a proposal are reasonable as a structural anchor, but confirm that OT-specific controls accompany them. A vendor whose methodology describes identify, protect and detect in IT terms but has no position on zone and conduit architecture, unidirectional gateways or Purdue model segmentation has not built an OT practice. Ask specifically about those topics during the evaluation call and note whether the answers come from experience or from a framework document.
OT domain expertise
Domain, weak process
Knows OT protocols but delivers inconsistent findings without a structured methodology.
The right choice
Deep OT knowledge combined with a structured, repeatable methodology and clear reporting.
Do not shortlist
Neither the domain knowledge nor the methodology required to run this engagement safely.
IT firm in disguise
Strong general methodology applied to the wrong question. Finds IT issues in OT wrappers.
Engagement methodology
OT and ICS security testing in the UAE
The UAE hosts oil and gas processing, electricity generation, water treatment, transport management and port logistics systems, many of which depend on industrial control networks in Dubai and Abu Dhabi. Federal entities and their suppliers operating supervisory systems fall under the UAE Information Assurance Regulation published by TDRA, which sets minimum security controls across information and operational technology alike.
DESC governs Dubai government entities and the suppliers who deliver technology services to them. An organisation providing operational support to a Dubai government entity is subject to DESC requirements that may include security testing of the operational systems in scope. In Abu Dhabi, entities in the healthcare sector are bound by ADHICS, and organisations in the energy and utilities sector work under the requirements of their licensing authority. CBUAE requirements apply to financial institutions, including those running operational technology for payment infrastructure.
Organisations procuring an OT/ICS pentest in either emirate should prepare a written authorisation naming the systems under test, the testing window, the methods permitted and the operational contact for the systems in scope. UAE regulators expect documented authorisation for security testing that affects operational systems. A verbal agreement or a general penetration testing mandate does not satisfy that requirement. The IT team and the operations staff who run the systems being tested should both review the authorisation letter, and a copy should be retained for any regulatory review.
Frequently asked questions
What is an OT/ICS pentest?
An OT/ICS pentest is a structured security assessment of industrial control systems, including PLCs, HMIs, SCADA systems and the networks connecting them. The tester looks for vulnerabilities that could allow an attacker to disrupt operations, manipulate process data or move laterally from the corporate network into the control environment. Unlike an IT pentest, the safety of live physical processes is a primary constraint on the test methods used.
How does an OT/ICS pentest differ from a standard IT penetration test?
The core methodology overlaps, but OT testing adds safety constraints, protocol-specific knowledge and operational considerations that do not apply to IT. OT systems often run legacy firmware that cannot be patched and use protocols with no built-in authentication. The consequences of a test error are measured in operational disruption rather than service availability. Passive reconnaissance, which captures and analyses traffic without sending commands, is far more common in OT work than active exploitation.
What does passive testing mean in an OT/ICS context?
Passive testing means the tester captures and analyses network traffic, device responses and configuration data without sending any commands to control devices. It can reveal unencrypted sessions, unauthenticated connections and protocol anomalies without putting live systems at risk. Most OT/ICS engagements begin with passive work and move to active testing only if it is approved, safely scoped and conducted in a maintenance window or against isolated hardware.
Which protocols does an OT/ICS pentest typically cover?
Common protocols include Modbus, DNP3, EtherNet/IP, Profinet, IEC 61850 and BACnet, depending on the industry and the age of the equipment. A vendor should name the protocols relevant to your environment and explain their approach to each. Protocols vary considerably in authentication and encryption support, and findings often turn on protocol-level weaknesses rather than software vulnerabilities.
Do I need to take systems offline for an OT/ICS pentest?
Not necessarily, but active testing against live PLCs or RTUs should be confined to maintenance windows or isolated test hardware where possible. Passive reconnaissance and perimeter testing can be conducted without operational impact in most environments. The scope and safety plan should specify exactly what will and will not be tested live, and the vendor should explain the risk profile of each test category before the engagement begins.
What qualifications should an OT/ICS pentest vendor hold?
Look for ICS-specific experience rather than general IT certifications alone. GIAC GICSP is a widely recognised OT security certification. More important is demonstrated experience: the industrial protocols the firm has tested, the industries they have worked in and a sample report from a comparable engagement. A vendor who holds CREST or OSCP for IT work but cannot name the protocols in your environment is not qualified for an OT engagement.
How long does an OT/ICS pentest typically take?
A scoped engagement focused on the IT-OT boundary and passive protocol analysis typically runs for one to two weeks of active testing, with preparation and reporting on either side. A broader assessment covering multiple zones, active device testing and detailed protocol analysis takes longer. The safety planning phase alone, involving coordination with operations staff and agreement of the authorisation document, often takes several days before testing begins.
What should a good OT/ICS pentest report contain?
A good report contains network topology diagrams of the control environment, findings described at the protocol or device level, evidence such as traffic captures or configuration extracts, and remediation advice specific to OT constraints. Recommendations should distinguish between vulnerabilities that can be patched, those that require compensating controls and those that need architectural changes. A report listing only CVSS scores and patch deadlines is not suitable for an OT environment.
Which UAE regulations require OT/ICS security testing?
The UAE Information Assurance Regulation applies to federal entities and their suppliers, including those operating supervisory systems. DESC requirements extend to Dubai government suppliers where operational technology forms part of service delivery. Sector regulators for energy, utilities and transport may impose additional requirements. Organisations in Abu Dhabi's healthcare sector are subject to ADHICS, which covers connected medical and building systems. Confirm which regimes apply to your entity before scoping the engagement.
How often should an OT/ICS pentest be conducted?
Annual testing is a reasonable baseline for environments where the architecture changes infrequently. An assessment should also follow any significant change to the OT network: adding a new control zone, connecting a previously isolated system to the corporate network, or deploying new HMIs or historians. After a suspected security incident, a focused assessment of the affected systems should be treated as a separate exercise rather than waiting for the next scheduled engagement.






