A chatbot should hand a customer over to a human agent when the customer explicitly asks for it, when the bot fails to understand the request after two attempts, or when the case is sensitive, complex or carries financial or legal implications. A good handover passes the full conversation context to the agent, so the customer never has to explain the problem again.

Situations that call for immediate transfer

  • An explicit request: if a customer asks to speak to a person, they should not be forced through extra steps or find the option hidden.
  • Repeated misunderstanding: if the bot fails to understand twice in a row, a third attempt usually adds to the frustration.
  • Complaints and emotional situations: language that signals annoyance or anxiety needs a human response.
  • Consequential decisions: disputing a charge, asking for a policy exception or anything touching on health or safety.
  • Priority customers: for example, a customer contacting you again about an issue that was not resolved before.

What a chatbot genuinely does well

Chatbots work efficiently on repetitive requests with clear answers: checking order status, opening hours, simple data updates, booking appointments and answering general policy questions. The more securely the bot is connected to the organization's systems, the more it can complete a request rather than just explain the steps. It is also important for the bot to identify itself clearly and say up front what it can do, so customer expectations stay realistic.

How to transfer without making the customer repeat themselves

  1. Pass the full transcript and a short summary of the request to the agent's screen.
  2. Carry over identity verification, if completed, and the initial request category.
  3. Route the conversation directly to the right team based on request type and the customer's language.
  4. Tell the customer clearly that they are being transferred, give an approximate wait time and offer a callback if the wait is long.
  5. Have the agent open by confirming what they understand of the request, not by asking the customer to start again.

Outside the human team's working hours, do not leave the customer at a dead end. Tell them clearly when agents will be available, create a logged request to be followed up at the start of the next shift and send them the reference number. Also make sure the number of agents assigned to receive transfers matches expected volumes at peak times, otherwise the handover itself becomes a new source of frustration.

Metrics that reveal problems

A high share of conversations closed by the bot alone is not enough to prove success. It may simply mean customers are giving up. Track more telling indicators:

  • The share of customers who come back about the same issue within a short period.
  • The share of conversations abandoned before resolution or before transfer.
  • CSAT for bot-only conversations compared with transferred ones.
  • Wait time after a transfer request, and the agent's average handling time afterwards.

Review a sample of transferred conversations each week to update the bot's flows and the knowledge base it relies on. RAM Orbit helps organizations design handover flows and run the human team that receives them under a single quality framework.

Key takeaways

  • Make the option to speak to a person clear and always available.
  • Transfer after two failed attempts and in any sensitive or consequential case.
  • Pass on the full context so customers do not have to repeat themselves.
  • Judge quality by repeat contact, abandonment and satisfaction, not by automation rate alone.

Let's Talk About Your Operations

Tell us about your operational needs, and we will work with you to assess your current setup and recommend the operating model that best fits your business.