A patient walks into an ER with chest pain. Another walks in with a sprained wrist. A third has a deep cut that's bleeding heavily. A fourth has a persistent cough.

The triage nurse sees all four within 90 seconds of arrival. Chest pain goes straight to a treatment room. The bleeding cut goes to urgent care. The sprained wrist gets an ice pack and waits. The cough sits in the waiting room.

Nobody processes them in the order they arrived. Nobody gives them equal attention. The system is explicitly unfair, and it saves lives because of it.

Your support queue should work the same way.

The Manchester Triage System

Most emergency departments worldwide use some version of the Manchester Triage System (MTS), developed in Manchester, England in 1996. It sorts patients into five levels:

Level 1 (Red, Immediate): Life-threatening. Seen now. No waiting.

Level 2 (Orange, Very Urgent): Potentially life-threatening or time-sensitive. Seen within 10 minutes.

Level 3 (Yellow, Urgent): Serious but stable. Seen within 60 minutes.

Level 4 (Green, Standard): Not urgent. Seen within 120 minutes.

Level 5 (Blue, Non-urgent): Minor issue. Seen when possible.

The key insight: the triage assessment takes under 60 seconds. The nurse doesn't diagnose the problem. They assess the severity using a structured decision tree: vital signs, pain level, mechanism of injury, and a few key questions. The diagnosis happens later, by a doctor. The triage just determines how fast the patient needs to be seen.

Applying This to Support

Most support teams have two priorities: "urgent" and "everything else." Some have no priority system at all. Tickets get processed in the order they arrive (FIFO), which means a billing error costing a customer $500 waits behind a "how do I change my email?" ticket that arrived 3 minutes earlier.

A support triage system works like the ER. Every incoming ticket gets a rapid severity assessment (under a second with AI, under 30 seconds for a human) and is routed to the appropriate priority queue.

Level 1 (Critical): Customer can't access their account, service is completely down for them, security breach, data loss. Response target: 15 minutes.

Level 2 (High): Billing error, broken critical feature, customer explicitly threatening to leave or chargeback. Response target: 1 hour.

Level 3 (Medium): Feature not working as expected, integration issue, moderate billing question. Response target: 4 hours.

Level 4 (Low): How-to questions, feature requests, general feedback. Response target: 24 hours.

Level 5 (Minimal): Spam, duplicate tickets, questions already answered in the same thread. Auto-close or batch process weekly.

The 60-Second Assessment

The ER triage nurse uses a structured decision tree. They don't think "what's wrong with this patient?" They think "does this patient need immediate care?" The questions are ordered by severity: is the patient breathing? Is there major hemorrhage? What's the pain level?

Your support triage follows the same logic. The questions aren't about the full solution. They're about urgency.

Does the customer mention money, billing, or charges? If yes, bump to Level 2 minimum. Financial issues create anxiety and chargeback risk.

Is the customer unable to use the product at all? If yes, Level 1 or 2. A customer who can't log in has zero value from your product right now.

Does the message contain legal, regulatory, or security keywords? If yes, Level 1. These have time-sensitive consequences.

Is the customer a high-value account? If yes, bump up one level. A $50K/year customer asking a how-to question still deserves faster attention than a free user with the same question.

AI classification handles this automatically. Supp classifies into 315 intents and assigns a priority score based on the intent type, message content, and urgency signals. The classification happens in under 200 milliseconds, faster than any triage nurse (though with lower stakes).

What ER Triage Gets Right That Support Doesn't

Reassessment. In the ER, triage categories aren't permanent. A patient classified as "urgent" who deteriorates gets reclassified as "immediate." A patient classified as "standard" whose vitals change gets reassessed.

Most support systems assign a priority once and never update it. A ticket classified as "low" that sits unanswered for 3 days should automatically escalate to "medium." A customer who sends a follow-up saying "this is now affecting our production environment" should get reclassified from medium to critical, instantly.

Build time-based escalation into your triage system. Level 4 tickets that aren't touched within 24 hours become Level 3. Level 3 tickets untouched for 8 hours become Level 2. The escalation ensures nothing rots in the queue.

Communication during the wait. ER waiting rooms have terrible reputations, but good ERs communicate: "You're next." "The doctor will see you in about 20 minutes." "We haven't forgotten about you."

Support should do the same. Customers in the Level 3 and Level 4 queues should get an acknowledgment with a realistic timeline: "We've received your message. You'll hear from us within 4 hours." That acknowledgment alone reduces anxiety and follow-up messages.

Separation of triage from treatment. The triage nurse doesn't treat the patient. They assess and route. The doctor treats. In support, the person who triages (or the AI that triages) shouldn't be the same person who resolves. Triage is about speed and accuracy. Resolution is about thoroughness and empathy. Different skills, different speeds.

The Uncomfortable Truth About Fairness

Triage is explicitly unfair. The person who arrived first doesn't get served first. The person with the most serious problem does.

This feels wrong to many support leaders. "Every customer deserves equal treatment" is a common value statement. But equal treatment doesn't mean equal timing. It means equal quality of resolution. The billing error customer and the how-to customer both deserve a great answer. But the billing error customer needs it faster because the consequences of delay are worse.

The ER doesn't apologize for treating chest pain before a sprained wrist. Your support team shouldn't apologize for prioritizing a billing crisis over a feature question. It's the right allocation of limited resources.

Supp's priority scoring does this math automatically. Every classified message gets a priority score based on intent urgency, customer value signals, and message content. The highest-priority tickets surface first. Your agents always work on the most impactful ticket available, not just the oldest one.

The ER figured out triage three decades ago. Support teams are still processing tickets in the order they arrive. The gap is enormous, and closing it is one of the simplest, highest-impact changes you can make to your support operation.