Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
An indicator of attack describes adversary behaviour rather than adversary infrastructure: what an intruder is doing right now, independent of which file, domain, or address they happen to be using.

What Is an Indicator of Attack?
An indicator of attack is a description of behaviour that signals an intrusion in progress. Where an indicator of compromise says “this file hash is malicious” or “this domain is a command-and-control server”, an indicator of attack says “a Word process has just spawned PowerShell with a base64-encoded command”. The first describes an artefact the adversary used. The second describes something the adversary did, and is far harder for them to avoid.
Artefact-based detection has an inherent lag. Someone has to observe the attack, extract the hash or the address, publish it, and let defenders ingest it — by which time the adversary has recompiled the binary and rotated the infrastructure, both of which take minutes. David Bianco’s Pyramid of Pain makes the point precisely: hashes and IP addresses are trivial for an attacker to change, while tactics, techniques, and procedures are expensive, because changing them means changing how the operation works.
Behavioural indicators sit at that expensive end. An intruder can swap tooling freely, but if their objective is to obtain credentials from a Windows host, something has to read the memory of the Local Security Authority process. The specific implementation varies; the behaviour persists.
Indicators of attack are what modern endpoint detection products actually run on, correlating process trees, memory access, file operations, and network connections rather than matching files against a list. In detection engineering they are expressed as rules — Sigma for portable, vendor-neutral logic, or a platform’s native query language — and mapped to techniques in MITRE ATT&CK so coverage can be measured honestly. A detection catalogue tagged by technique shows which parts of the kill chain you can see and, more usefully, which you cannot.
The same thinking drives threat hunting. A hunt starts from a hypothesis about behaviour, not from a feed of addresses: if an adversary were using scheduled tasks for persistence in this estate, what would that look like in the data I hold?
Behavioural detection generates false positives, and anyone claiming otherwise is selling something. Administrators use PowerShell. Backup software reads unusual files. Deployment tooling installs remote services at scale. The techniques that make this workable are environmental baselining, correlating several weak signals into one strong one, and tuning per environment rather than globally. Expect to invest continuing effort: a rule written in January and never reviewed will be either noisy or silent by June.
The practical answer is to run both kinds of indicator. Artefact matching is cheap, exact, and perfect for sweeping an estate for a known web shell. Behavioural detection catches the intrusion for which no indicator has been published yet. Organisations that rely on one alone either drown in alerts or discover intrusions through a third party.
At Privia Security, we build and tune behavioural detections against real adversary techniques through our SOC maturation services, and validate them with red team activity rather than assuming the rule fires.
An indicator of compromise is an artefact associated with a known attack: a file hash, a domain, an IP address, a registry key. An indicator of attack describes behaviour, such as a document spawning a shell or credential material being read from memory. Artefacts change in minutes; behaviours persist across campaigns, which is why both belong in a detection strategy.
No, they complement it. Signature and hash matching is cheap and exact, which makes it ideal for sweeping an estate for a known tool or blocking previously seen malware. Behavioural detection covers the novel case where no artefact has been published. Removing either one leaves a gap that adversaries exploit routinely.
Baseline the environment first, so you know which administrative tools run where and under which accounts. Correlate multiple weak signals rather than alerting on a single one, scope rules to the systems where the behaviour is genuinely abnormal, and review every rule on a schedule. Untuned behavioural detection produces alert fatigue faster than any other source.
Yes, and you should. Sigma provides a vendor-neutral rule format that converts to most SIEM query languages, and public rule repositories give a starting library mapped to ATT&CK techniques. The most valuable rules are usually those written for your own environment, because they encode knowledge about what normal looks like that no external feed has.
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.