Most SIEM deployments are good at recognizing what has been seen before. Known signatures, malicious hashes and blocklisted addresses generate alerts reliably. The harder problem is the activity that comes before a compromise is obvious. This article explains how to build SIEM behavioral detection rules for the threat patterns that matter in Saudi regulated environments, how to baseline normal administrative activity, and how to prove the result during a security operations assessment.
Why behavior matters more than signatures
An intruder rarely announces itself with a known payload. It tests credentials, maps the network, gathers data and prepares a route out. Each step may look ordinary in isolation. Detection that depends on signatures sees none of this, while detection that models behavior can connect the steps into a sequence an analyst can act on.
For entities working under the NCA Essential Cybersecurity Controls (ECC), the expectation is not only that logs are collected but that monitoring is effective and that you can show it. Evidence of tested detection is far stronger than evidence of rules merely being switched on.
Five behavioral patterns to model
1. Credential enumeration
Look for many failed authentications across multiple accounts from one source, one account failing from many sources, and repeated queries against directory services. The useful signal is the spread and tempo, not any single failure.
2. Lateral-movement preparation
Before moving, an actor discovers what is reachable. Model new administrative protocol use between hosts that never communicate, remote service creation, and a workstation suddenly contacting many internal systems.
3. Privilege-escalation attempts
Correlate additions to privileged groups, unusual use of administrative tools by standard accounts, and sensitive-role assignments outside change windows. Pair these with the change records so approved work is distinguished from the rest.
4. Data staging
Staging often appears as large file collection, archive creation or copying to an unusual internal location, typically from an account that does not normally handle that data.
5. Exfiltration setup
Watch for new outbound destinations, first-time use of external storage or file-transfer services, and volume or timing that departs from the host's history.
The value comes from chaining these. A rule that fires on enumeration followed by escalation on the same entity within a defined window is more useful than five unrelated low-severity alerts.
Baselining normal administrative activity
False positives are the main reason behavioral rules get disabled. Baseline before you tune.
- Inventory privileged accounts, administrative jump hosts and scheduled maintenance windows.
- Collect a representative period of activity so routine patching, backups and vulnerability scanning are visible as normal.
- Register authorized scanners and management tools by address and account so their activity is labeled, not ignored.
- Link rules to change records where possible, so approved work is distinguished from unapproved work.
- Review thresholds after each tuning cycle and record the reason for every change.
Document the baseline. An assessor will ask how you decided what normal looks like.
Three alert-validation scenarios
Validation turns a rule from an assumption into evidence. Agree each scenario in advance with system owners, define the scope and a stop condition, and notify the people who need to know.
Scenario 1: Simulated reconnaissance
Run a controlled scan and directory-query sequence from a test host. The goal is to show that the SOC distinguishes it from the authorized vulnerability scanner already in the baseline. Record whether an alert fired, how it was triaged, and whether the analyst correctly classified the registered scanner as benign and the test host as suspicious.
Scenario 2: Controlled lateral movement
From a test account, authenticate to a small set of systems it has never accessed, using an administrative protocol. Confirm that the correlation links the discovery, the new connections and any escalation into one case, rather than scattering them across separate alerts.
Scenario 3: Staged data transfer
Collect a test dataset, compress it, and send it to an approved external test destination. Confirm that staging and outbound transfer are both detected, and measure the time from activity to analyst awareness.
For each scenario, keep the plan, the timestamps, the alert records, the analyst notes and any rule changes that followed. This package is what supports your evidence for NCA ECC monitoring and incident-management expectations during an assessment.
What a good outcome looks like
A mature result is not a large rule count. It is a documented set of detection use cases tied to realistic threats and to your regulatory obligations, a baseline that explains normal activity, validation records showing the rules work, and a clear path for the gaps found. Analysts should know what each alert means and what to do next.
Signs it is time to review
- Alerts are dominated by signature matches, with little correlation across events.
- Analysts routinely close scanner and administrator activity as noise without a documented baseline.
- No one can show when a detection rule was last tested.
- An assessment or audit is approaching and evidence of monitoring effectiveness is thin.
- A managed service provider reports on volumes, not on detection quality.
How we approach it
Our Security Operations Advisory service covers the design, assessment and improvement of security operations, monitoring and incident response capabilities. The work typically runs 4–10 weeks depending on scope. Deliverables include a security operations maturity assessment, a detection use-case catalog, incident response plan and playbooks, a tabletop exercise with lessons learned, and an improvement roadmap.
In practice this means a measured view of your detection and response maturity, and detection use cases prioritized by threat and obligation. Where you rely on a provider, we review managed service contracts against requirements. Our role is advisory: we design, assess and improve security operations and help you select and oversee providers, so our advice stays independent.
Next step
If your SIEM mostly detects what is already known and an assessment is coming, start with the five patterns and the three scenarios above. If you would like an independent view of where your detection stands, our Security Operations Advisory team can assess it and set out a prioritized roadmap.