Short answer
Escalation in customer support is passing a request to someone with more expertise or authority when the current handler cannot resolve it, such as a specialist team, a manager or engineering. Good escalations carry the full context, so the customer never explains twice.- Also called
- Support escalation, ticket escalation, an escalated ticket
- Two main types
- Functional, to more expertise; hierarchical, to more authority
- Common triggers
- Out of scope, needs approval, upset customer, deadline at risk
- Measured by
- Escalation rate, time to escalate, outcome after escalation
Every support team has requests the first person cannot finish: a refund above their limit, a bug only engineering can fix, a customer who wants a manager. Escalation is how those requests move on. Done well, it is invisible to the customer. Done badly, it is the moment they repeat their story for the third time.
Types of escalation
Most escalations are one of three kinds.
- Functional escalation. The request moves sideways to a team with the right expertise, such as billing, logistics or engineering.
- Hierarchical escalation. It moves up to someone with more authority, such as a team lead approving a refund above policy.
- Automatic escalation. A rule moves it without anyone deciding, for example when a response deadline is about to be missed.
An AI agent handing a conversation to a person is an escalation too, and follows the same rules.
When to escalate
Escalate when finishing the request needs something the current handler does not have.
- The request is outside their scope or knowledge.
- It needs an approval they cannot give, such as a refund above their limit.
- The customer asks for a person or a manager.
- The customer mentions a chargeback, a lawyer or a safety problem.
- A deadline is at risk, such as an order about to ship to the wrong address.
- The same issue has already come back once.
How an escalation process works
A clear process keeps escalations fast and stops them bouncing between people.
- Recognize the trigger, using rules everyone knows.
- Choose the owner: a named team or person, not "someone senior".
- Pass the context: the order, what the customer wants, what was already tried.
- Tell the customer who is taking over and when they will hear back.
- Track the escalation until it is closed, with a deadline.
- Close the loop with the person who escalated, so they learn the answer.
With Aide, a procedure that ends in a handoff can first collect the order number and the reason for contact, one question at a time, and assign the conversation to a named group. The owner is set and the context is on the conversation before anyone opens it.
Escalation examples in a Shopify store
Four requests that should leave the front line, and where they go:
- "The vase arrived in pieces, I want a full refund and to keep the replacement." A refund above the agent's limit goes to a team lead.
- "I'll dispute this with my bank if you don't refund today." A chargeback threat goes to whoever handles payments.
- "My order shipped to my old address yesterday." The change window has closed, so it goes to the person who manages the carrier.
- "Your product gave me a rash." A safety complaint goes to a manager, the same day.
In Aide, the first example escalates by design: refunds wait for the team by default, and a large refund from a teammate with the Agent role waits for an Owner or Admin, who gets an "Order action needs approval" notification and approves or rejects it on the order.
Escalation analysis
Escalation analysis means reviewing escalated conversations by reason and by request type to find causes that can be fixed. A cluster of escalations about one policy usually means the policy, or the help article about it, needs to change. The escalation rate on its own says little; the reasons behind it say what to do.
The practice this page rejects is treating every escalation as a failure. An agent, human or AI, who never escalates is guessing on the requests they should hand over.
Where Aide fits
When Aide hands a conversation to the team, it leaves a handoff note in labelled sections and can add its draft reply above the summary, so the person picking it up starts with the context. Teams write hand-off rules in plain English, such as "the customer threatens legal action", and a shopper who clicks Talk to a person reaches the team at once. Aide's analytics break handoffs down by reason and channel. See Aide Helpdesk.