When should AI customer service escalate to a human?
The situations where AI customer service should hand over to a person, how to design escalation rules, and how to make the handover smooth for the customer.
The fastest way to lose a customer’s trust is to trap them with an AI that won’t let them reach a person. The second fastest is to hand them to a person who asks them to start again. Getting AI customer service escalation right, deciding when a conversation should move from AI to a human and making that move smoothly, is what separates AI that customers tolerate from AI that genuinely helps.
This article sets out the situations where AI should hand over, how to write escalation rules for your business, and how to design the handover itself. The principles apply whether you run a property management agency, a trades business or any other service business.
Why escalation matters more than automation
Customers are cautious about AI in customer service, and their main fear is specific. In a Gartner survey of 5,728 customers, 64% said they would prefer companies didn’t use AI for customer service. The top concern was that AI would make it harder to reach a person, and 53% said they would consider switching to a competitor if they found out a company was going to use AI for customer service (Gartner, 2024).
That finding is often read as a reason not to use AI. It’s better read as a design brief. If customers fear being stuck, the answer is an AI that makes reaching a person easy, says clearly that it’s AI, and hands over with the conversation intact. Escalation isn’t a failure of the AI. It’s part of how good AI customer service works.
Seven situations where AI should hand over
1. The customer asks for a person
This is the simplest rule and the most important. When someone asks for a person, the AI should say who will help and how, then hand over. Arguing, asking them to rephrase or offering more AI help first is what customers hate most.
2. The AI isn’t confident it understands
If the AI can’t work out what the customer needs after a reasonable attempt, it should stop guessing. A brief “Let me get someone who can help with this” is better than a loop of clarifying questions.
3. The conversation is going in circles
Repeated questions, the same answer given twice or a customer rephrasing the same request are signs the conversation isn’t working. Setting a limit, such as two failed attempts, prevents customers getting stuck.
4. The customer is upset
Frustration, anger or distress call for a person. AI can recognise signs of frustration, but calming someone down and rebuilding goodwill is human work. An upset tenant, a customer whose job went wrong or someone making a complaint should reach a person quickly.
5. The topic is sensitive or high-stakes
Refunds, disputes, billing problems, legal questions, complaints and anything involving a customer’s personal circumstances should go to a person, even if the AI could technically respond. These conversations carry risk, and they deserve judgement.
6. There’s a safety issue
Anything involving risk to people or property, such as a gas smell, flooding or a security issue, needs a person quickly. Decide in advance what the AI tells the customer to do first, for example to contact emergency services, and who it alerts at the same time.
7. The answer requires a decision
Approving a cost, making an exception to a policy or agreeing to something not covered by your rules are decisions. The AI can gather the information and set up the decision, but a person should make it.
Writing escalation rules for your business
Generic rules are a starting point, but every business has its own sensitive topics, emergencies and decisions. The practical approach is to list the situations your team handles personally today, decide which of them must always reach a person, and write them down as plain rules: what triggers the handover, who receives it, how quickly and through which channel. A property management agency might route all disputes about bond deductions to the property manager, while a trades business might route every complaint about completed work to the owner. Our article on why AI customer service fails shows how unclear handovers cause many of the problems blamed on AI.
Then decide what happens outside business hours. If nobody is available to take a non-urgent handover at 10pm, the AI should say so honestly, capture the details, tell the customer when they’ll hear back and make sure someone does follow up.
Designing a smooth handover
The handover is where customers judge the whole experience. Three things make it work. The person receiving the conversation should get the full history, a short summary and the reason for the handover, so the customer never has to repeat themselves. The customer should be told what’s happening and roughly when, for example “I’ve passed this to our team and they’ll call you within the hour”. And someone must own the follow-up, so a handover never becomes a message that sits unread.
- The full conversation Everything the customer has already said.
- A short summary What they need, in one or two lines.
- The reason Why the AI handed over.
- A clear next step What the customer was told, and when.
- An owner The person responsible for following up.
Reviewing escalations over time
Escalation rules shouldn’t be set once and forgotten. Reviewing escalations regularly shows whether the AI is handing over too early, which wastes your team’s time on routine questions, or too late, which frustrates customers who needed a person sooner. Patterns in escalations often reveal gaps in the AI’s information or unclear policies that the business can fix. Our guide to 15 ways to use AI agents for customer support covers which tasks typically suit AI and which don’t.
How Shift AI hands conversations to the right person
Escalation is built into how Shift AI works, rather than added on. Every query is answered, sorted and followed through, and you decide which parts our AI handles and which go to people.
Run: AI agents set up around your portfolio
Our AI agents answer calls, chats, texts and emails, always say they’re AI and follow your escalation rules: what triggers a handover, who receives it and how fast. When a customer asks for a person, they’re handed over, no argument. Urgent and safety issues alert the right person straight away.
Scale: people for the conversations that need judgement
The people receiving the handover can be your own team, who can join any conversation in the Shift app with the full history in front of them, or a Shift specialist we manage, who takes the conversations that need judgement and follows them through. Either way, the customer doesn’t start again.
Management: reviewed and improved every month
Each month we review escalations with you: how many there were, why they happened and how quickly they were picked up. Your scorecard shows the trend, and we adjust the rules and the AI’s information so it hands over at the right moments, not too early and not too late.
In practice, a customer chatting about a booking change who then mentions a complaint about the last visit is answered on the booking, while the complaint is passed to your team with a summary and the customer is told someone will call them that afternoon. The team member sees exactly what was said and calls with the context already in hand.
If you want AI that answers quickly without trapping your customers, book a free AI strategy sprint and we’ll map which conversations should stay with AI and which should always reach a person.