Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
User and entity behaviour analytics builds a model of what is normal for each account and device, then scores deviations — catching misuse that no signature or rule describes.

What Is UEBA?
Rules describe attacks somebody has already seen. An account that signs in at its usual hour, from its usual country, using its usual device, and then downloads the entire customer database breaks no rule at all. User and entity behaviour analytics exists for that case: rather than describing malicious activity, it builds a statistical model of normal activity for each user and each device, and flags behaviour that departs from it.
The term grew out of user behaviour analytics, with analysts at Gartner adding the “entity” in 2015 to reflect that servers, service accounts, and devices need modelling too — and that service accounts, which behave with machine-like regularity, are among the easiest things to baseline and the most damaging to have stolen.
Three mechanisms do most of the work.
Individual baselining. For each account, the system learns working hours, usual locations and devices, the applications touched, the volume of data normally accessed, and the systems normally authenticated against. Deviations are scored by how far they fall outside that distribution.
Peer group comparison. An individual baseline misses the employee who has always behaved unusually. Comparing against colleagues in the same role, drawn from HR and directory data, catches the finance analyst accessing engineering repositories that no other finance analyst touches.
Risk aggregation. This is the part that distinguishes UEBA from simple anomaly detection. No single weak signal is worth an alert — an unfamiliar location, a first use of a new application, a modest increase in downloads. Accumulated against one identity within a short window, they produce a risk score that crosses a threshold, and the alert describes a sequence rather than an event.
Cold start. Baselines need weeks of data, typically two to four, before scores mean anything. Deployments evaluated in the first fortnight look terrible and are judged unfairly.
Identity data quality. Peer grouping depends on accurate role, department, and manager information. Where the directory is stale and half the organisation is in a group called “Users”, peer comparison produces noise. The preparatory work is identity hygiene, not analytics.
Legitimate change. Promotions, project moves, secondments, acquisitions, month-end and year-end processing, and a sudden shift to remote working all cause genuine behavioural change that scores as anomalous. Feeding HR change data into the system helps; nothing removes the problem entirely.
Explainability. An analyst handed “risk score 87” with no breakdown cannot act. Insist on seeing which contributing behaviours produced a score, or triage becomes guesswork.
Slow adversaries. Behaviour that changes gradually over months can be absorbed into the baseline, which is exactly what a patient insider produces.
Most organisations now meet UEBA as a feature of their SIEM or identity platform rather than as a separate product, which is sensible: the analytics are only as good as the telemetry feeding them, and the same data serves both purposes.
UEBA profiles identifiable employees, which makes it one of the more sensitive technologies a UK security team deploys. Processing is normally justified under legitimate interests, but UK GDPR requires that interest to be balanced against the rights of the individuals monitored, and the ICO’s guidance on monitoring workers is explicit that covert or excessive monitoring will not be defensible.
In practice that means completing a data protection impact assessment before deployment, telling staff what is monitored and why in language they will actually read, restricting access to behavioural profiles to named investigators, logging who queries them, and keeping a human decision between a risk score and any action affecting someone’s employment. Automated decisions with legal or similarly significant effects attract additional restrictions under Article 22. Consulting employee representatives early tends to produce a better outcome than defending the system afterwards.
At Privia Security, we help organisations deploy and tune behavioural analytics proportionately through our SOC maturation services.
Anomaly detection is one component. UEBA adds identity context, peer group comparison, and risk aggregation across many weak signals attributed to the same user or entity over time. A standalone anomaly detector flags an unusual event; UEBA builds a case that a specific identity is behaving in a way that warrants investigation, which is far more actionable.
Expect two to four weeks of baselining before scores are meaningful, and two to three months before tuning settles. Plan the evaluation accordingly: judging a deployment in its first fortnight will produce a misleading verdict. Volume of false positives should fall noticeably as peer groups are corrected and legitimate seasonal patterns are learned.
Yes. UK GDPR transparency obligations and the ICO’s guidance on monitoring workers both expect clear information about what is monitored, why, and on what lawful basis, normally through a privacy notice and an acceptable use policy. Covert monitoring is only justifiable in narrow circumstances, usually where there is suspicion of serious wrongdoing and notification would prejudice an investigation.
No, and vendors claiming otherwise should be questioned. Rules catch known attack patterns precisely and cheaply, and they remain the backbone of detection. UEBA covers what rules cannot describe: misuse of legitimate access, and activity that is only suspicious relative to a particular user. The two are complementary, and most real detections come from the rules.
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.