Explore our new Cybersecurity Terms Glossary, explained by our experts.
Cybersecurity Terms Glossary
An alert prioritisation model is the agreed, written method a SOC uses to decide which alert is worked first, combining detection confidence, asset value, identity privilege, and observed stage of attack.

What Is an Alert Prioritisation Model?
Every security team prioritises. The question is whether the method is written down and applied consistently, or whether it lives in the head of whoever is on shift. An alert prioritisation model makes it explicit: a defined set of inputs, a defined way of combining them, and a resulting order of work that two different analysts would arrive at independently.
Detection products ship a severity field, and it describes the rule rather than your organisation. The same “suspicious PowerShell execution” alert is a formality on a developer’s workstation during working hours and a serious matter on a domain controller at 03:00. A model that ignores the target cannot distinguish the two, which is why queues sorted purely on vendor severity drift towards being sorted by nothing at all.
Keep the arithmetic simple enough that analysts can reproduce it mentally. One workable scheme multiplies confidence by impact, where impact is the higher of asset criticality and identity privilege, then applies a stage multiplier:
| Input | Scale | Example |
|---|---|---|
| Confidence | 1–5 from measured precision | Rule with 70% true positives scores 4 |
| Impact | 1–5 from asset tier or privilege | Domain controller scores 5 |
| Stage multiplier | 0.8 to 1.5 by ATT&CK tactic | Credential access scores 1.3 |
A confidence of 4 on a tier-5 asset at credential access gives 26, which sits well above a confidence of 5 on a tier-2 asset at discovery, which gives 8. The numbers matter less than the fact that the ordering is defensible and that the same alert tomorrow produces the same answer.
Publish the model and the bands it maps to, so a score of 20 or more means work it now and a score below 6 means close it in bulk unless something else is attached to the entity. Give analysts an explicit override, and require a one-line reason, because the overrides are the richest feedback you will get on where the model is wrong.
Feed the model from enrichment automation rather than from manual lookups; a scoring scheme that requires three searches per alert will not survive a busy Monday. And review the confidence values quarterly against actual outcomes, since a rule’s precision drifts as the environment changes.
Stale asset data is the most common. If the inventory still lists a decommissioned server as tier one and omits the new payment gateway, the model confidently produces the wrong order. Treat inventory freshness as part of the detection capability, not as someone else’s problem.
Score inflation is next. When every system is critical, nothing is, and the model collapses back to vendor severity. Cap the proportion of assets that may hold the top tier, typically under 5%, and make the business justify additions.
Finally, remember that prioritisation orders the queue; it does not shorten it. If 60% of alerts score low enough to be closed without review, that is evidence for rule tuning, not a sign the model is succeeding.
At Privia Security, we usually start a prioritisation project with the asset tiering rather than the scoring formula, because the formula is the easy half. Our SOC maturation services cover both.
Prioritisation decides the order in which alerts are picked up; triage is the assessment performed once one is picked up. Prioritisation happens automatically against every alert in the queue, usually within seconds of creation. Triage is manual, takes minutes, and ends with a classification and either closure or escalation. A good model means analysts spend their triage time on the alerts most likely to matter.
The scoring can and should be. The decision about what to do with the resulting order should not be fully automated below the highest confidence bands, because scoring depends on data that goes stale and on assumptions that attackers can learn. Automate the arithmetic, automate the enrichment that feeds it, and keep a human deciding what happens to anything that scores above the action threshold.
Start with a partial one. Listing your domain controllers, internet-facing systems, payment and finance applications, and the top 20 servers by business dependency gives you most of the value within a fortnight. Everything else defaults to a middle tier. A partial inventory applied consistently beats a complete one that never arrives, and the gaps become an obvious piece of follow-up work.
Yes, but as a modifier rather than a primary input. If credible reporting says a group active against your industry is using a specific technique this month, temporarily raising the confidence weighting for detections mapped to that technique is sensible. Set an expiry when you apply it, otherwise the model accumulates permanent adjustments that nobody remembers the reason for.
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.