Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
Detection engineering treats security detections as software: written against requirements, held in version control, tested before release, measured in production, and retired when they stop earning their place.

What Is Detection Engineering?
For most of the last two decades, security detections were written by whoever had access to the console, saved inside the product, and forgotten. Detection engineering is the response to that: a discipline that applies ordinary software practice to detection content, so that rules have requirements, authorship, tests, versions, and a measurable production record.
A detection engineer spends their time on a loop with five parts. Research: understanding how a technique actually works, usually by executing it in a lab and watching what telemetry it produces. Requirements: establishing which data source carries the evidence and whether the organisation collects it. Development: writing logic that is specific enough to be actionable and general enough to survive variation. Validation: proving it fires against a real execution and establishing its false positive profile. Operation: watching how it behaves in production and feeding that back.
The step most often skipped is the first. Rules written from a blog post without executing the technique tend to detect the specific command the blog used, which is why so many estates have detections for -EncodedCommand and nothing for the same payload delivered through a different switch.
The practice that defines the discipline is keeping detections in a repository rather than in a product. The benefits are the ordinary ones: review before deployment, a history explaining why a condition exists, the ability to roll back a change that caused an alert storm, and automated testing in a pipeline.
Sigma makes this portable. A rule written once in its YAML format can be converted to Splunk, Microsoft Sentinel, Elastic, and others through pySigma backends, which keeps the logic independent of the platform you happen to be renting this year. Most teams keep a repository containing the Sigma rule, a test case built from captured telemetry, triage guidance, and an ATT&CK mapping, then deploy from the pipeline into the production platform.
Documenting each rule in the style of Palantir’s Alerting and Detection Strategy framework keeps the reasoning with the code: the goal, the categorisation, the technical context, the blind spots, the known false positives, the validation steps, and the response plan. The blind spots section is the part that protects your successor.
Analysts consume alerts; engineers produce them. The skills overlap but the orientation is different, and the most effective arrangements keep a deliberate feedback path between the two. An analyst closing the same benign case for the fourth time should be able to raise it as a defect rather than simply closing it faster, and the engineer should see the triage outcomes for every rule they own.
Rotating analysts into engineering work for a period is the cheapest way to build the capability and tends to produce better rules than hiring engineers who have never worked a queue.
Counting rules written is the metric to avoid. It rewards volume, and volume is what produces the queue described under alert fatigue.
Smaller organisations can adopt the practice without a dedicated function. Put your existing rules in a repository, even if deployment stays manual. Write a short document for each one covering what it detects and what to do when it fires. Record triage outcomes consistently enough to calculate precision. Pick the five techniques most relevant to your sector and build tested detections for them properly rather than importing 300 community rules.
That sequence delivers most of the benefit, and it is where we start with clients in our SOC maturation services, because the discipline matters considerably more than the tooling.
A working knowledge of attacker techniques, fluency in the query language of your platform, enough systems administration to understand what normal looks like on Windows, Linux, and your cloud estate, and the software habits of version control, testing, and documentation. Experience as a SOC analyst is the most useful background, because it makes the difference between a rule that is technically correct and one that is actually workable.
No, but it helps. You can keep platform-native rules in version control and get most of the review and history benefits. Sigma adds portability, which matters when you change SIEM, run more than one platform, or want to consume community rules without rewriting them. The cost is an extra conversion step and occasional loss of platform-specific features in complex logic.
Fewer than most have. A few hundred well-tested, owned, and tuned rules outperform several thousand imported ones, because the imported majority generate noise that hides the valuable minority. Judge the estate by measured precision and validated coverage rather than by rule count, and make retiring unproductive rules a routine activity rather than an exceptional one.
It usually works best adjacent to the SOC rather than inside the alert queue, because engineering work is interrupted continuously if the same people are on shift. Keep a formal feedback loop in both directions: engineers see triage outcomes for the rules they own, and analysts can raise rule defects as work items. Rotating staff between the two roles builds both capabilities.
You May Be Interested In These
We use a few cookies that are needed to make this website work, and they are always on. We would also like to set analytics cookies, to see which pages are useful; functional cookies, so that our live chat works; and advertising cookies, to measure our campaigns. We will not set any of those until you say yes, and you can change your mind at any time from the Cookie Settings link in the footer.
Choose the cookies you are happy for us to use. Strictly necessary cookies are always on; everything else is off until you turn it on.
These cookies keep the website working: they hold your security checks, your form submissions and the choice you make here. The site cannot run without them, so they are always on.
These let our live chat open and remember your conversation, and let the newsletter sign-up work. Switch them off and the chat window will not appear.
These tell us which pages are read and how people find us, so that we can improve the site. The reports are about groups of visitors, never about you personally.
These let us measure our advertising and show you our adverts on other websites, such as LinkedIn, Google and Meta. Switch them off and you will still see adverts, but they will be less relevant.