Client Handoff Playbook (Handoff)

Client Handoff Playbook (Handoff)

1. 01

   Freeze the automation scope in a signed one-page handoff sheet before any client login changes hands

   List every AI-resolved intent, every escalation path, and the exact channels (voice, SMS, chat, email) in scope. Ambiguity at handoff is the most common cause of month-two disputes on a managed retainer.

2. 02

   Document who controls the meter, the gate, and the referee for each platform in the stack

   Forrester flags that AI application vendors now set usage limits, control feature access, and arbitrate their own disputes. Name the account owner, the billing contact, and the escalation route for each vendor so the client is not discovering this during an outage.

3. 03

   Replay 20 recorded interactions with the client in the room, split across resolved and escalated calls

   Include at least three edge cases: a multilingual caller, a caller who interrupts the bot, and a caller requesting a human. Agreement on what 'good' sounds like beats any written acceptance criteria.

4. 04

   Transfer ownership of the training corpus and intent taxonomy, not just the login

   Export the phrase lists, vertical vocabulary, and routing rules to a client-owned repository. A platform like Glassix or Talkdesk will hold this data, but the client should be able to walk it to another vendor without a rebuild.

5. 05

   Set the human oversight ratio in writing and attach it to the retainer line item

   State how many live-agent hours per 100 AI-handled conversations the agency will staff, and what triggers a renegotiation. This is the differentiator that keeps the service from being priced as a commodity seat license.

6. 06

   Run a 14-day parallel period where the agency keeps shadow access after the client takes the wheel

   Shadow access is read-only and time-boxed. It catches configuration drift in week one, which is when most handoffs quietly fail.

7. 07

   Schedule the 30-day review with a fixed agenda: containment rate, escalation accuracy, and cost per resolved contact

   Bring the baseline numbers captured at kickoff so the comparison is against the client's own prior performance, not an industry benchmark they cannot verify.