CHATBOT-TO-HUMAN HANDOFF CHECKLIST AND AVAILABILITY TEST RECORD
By Roman Mohren | FullSession | Prepared 19 October 2026
A reusable working tool. Every observation field is blank on purpose: fill it in from your own setup. Nothing here reports a completed test.

HOW TO USE THIS FILE
Part A is the five-field handoff record. Use it for each handoff you review, or paste it into an internal note when a person takes over.
Part B lists the triggers to define and test.
Part C is the availability test for an unavailable teammate. Run it before launch and again after you change office hours, the away message, permissions or staffing.
Part D is a weekly review sheet.
Expected results in Part C describe FullSession Desk behavior as published on fullsession.io/solutions/support-sales-desk, checked 19 October 2026. Record what you actually observe.


PART A: THE FIVE-FIELD HANDOFF RECORD

Conversation reference:
Date and time of handoff (with time zone):
Trigger (see Part B):

1. THE QUESTION
In the customer's own words:
What the customer is trying to get done:

2. ATTEMPTED STEPS
What the chatbot answered:
Source behind each answer (article, page or policy and its version):
What the customer already tried:
State of any attempted change (circle one): completed / failed / unknown / none attempted
If unknown, how the state will be checked before anyone retries:

3. AVAILABLE EVIDENCE
Transcript available in the same thread? (yes / no):
Account context available (plan, role, workspace or order), limited to what your connection permits:
Page or screen where the problem happened:
Session recording (only with a capture product such as Core CX / UX, and only if this session was captured): captured / not captured / not used
What the recording showed, if opened:
Sensitive data kept out of notes and attachments? (yes / no):

4. NEXT OWNER
Assigned person or team:
Who else was @mentioned or informed, and why:

5. PROMISED NEXT UPDATE
What the customer was told, word for word:
Time of the promised update (with time zone):
Where the update will appear (same thread / other channel, tested):
Update delivered as promised? (yes / no, and when):

Receiving teammate's check:
Did I have to ask the customer for anything already in the thread? (yes / no):
If yes, what was missing:


PART B: HANDOFF TRIGGERS TO DEFINE AND TEST
Write each trigger as a rule you can test. Run each one with a real question from your own queue.

Trigger 1. The customer asks for a person
Rule: hand over at once, without making the customer rephrase or argue.
Test question used:
Observed result:
Date tested:

Trigger 2. Unsupported or unpermitted action
Rule: explain the limit once, then hand over with the requested change written down.
Test question used:
Observed result:
Date tested:

Trigger 3. Account or billing exception
Rule: route to someone who can approve it; the chatbot does not improvise policy.
Test question used:
Observed result:
Date tested:

Trigger 4. Low confidence or repeated failure
Rule: no approved source, or the problem persists after two attempts: stop guessing and hand over with the open question.
Test question used:
Observed result:
Date tested:

Trigger 5. Policy-sensitive topic
Rule: security, data deletion, legal or contract questions go to a person wherever your policy requires it.
Test question used:
Observed result:
Date tested:

Trigger 6. Frustration
Rule: acknowledge it and offer a person rather than another answer.
Test question used:
Observed result:
Date tested:


PART C: AVAILABILITY TEST RECORD (UNAVAILABLE TEAMMATE)

Tester:
Date:
Site or app tested:
Device and browser used for the customer side:
Office hours configured (days, hours, time zone):
Away message, word for word:

Step 1. Configure
Do: set office hours in your team's time zone and write the away message.
Expect: hours and message saved as written.
Observed:
Pass / fail / unknown:
Notes:

Step 2. Ask for a person after hours
Do: outside those hours, open the chat as a visitor (or as a test customer in your app) and ask for a person.
Expect: the away message appears. Nothing promises an instant reply.
Observed (exact text shown to the customer):
Pass / fail / unknown:
Notes:

Step 3. Leave a question
Do: leave a specific question in the conversation, for example a setting that will not save.
Expect: the question is accepted and reaches the shared inbox.
Observed:
Time the question appeared in the inbox:
Pass / fail / unknown:
Notes:

Step 4. Reply the next business day
Do: a teammate assigns the conversation and replies in the same thread, stating a next update time.
Expect: the assignment is visible and the reply is in the original thread.
Observed:
Assigned to:
Time of first human reply:
Pass / fail / unknown:
Notes:

Step 5. Return as the customer
Do: come back as the same visitor (or as a test customer in your app); record the device and browser.
Expect: the same thread, with the reply, and nothing to restate.
Observed:
Did the customer have to restate anything? (yes / no):
Pass / fail / unknown:
Notes:

Step 6. Recording, if used
Do: with Core CX / UX, check whether the customer's session was captured and whether the teammate could open it from the conversation.
Expect: a recording only when the session was captured under your plan, capture settings and retention.
Observed (captured / not captured / not used):
Pass / fail / unknown:
Notes:

Step 7. Any channel your away message mentions
Do: if the away message promises another channel (for example email), check that a reply actually reaches the customer there.
Expect: only promise a channel you have observed working in your own setup.
Observed:
Pass / fail / unknown / not applicable:
Notes:

Decision after the test:
Changes needed to the setup or to the promise before launch:
Owner of each change:
Retest date:


PART D: WEEKLY HANDOFF REVIEW
Week of:
Conversations the chatbot handled:
Conversations handed to a person:
Handoff rate by trigger (handed over for that trigger, divided by conversations handled):
  Trigger 1:
  Trigger 2:
  Trigger 3:
  Trigger 4:
  Trigger 5:
  Trigger 6:
Time to first human reply, office hours (median):
Time to first human reply, after hours (median):
Context-sufficient handoffs (count of "no, I did not have to ask" answers) and sample size:
Reopen or repeat contact within seven days:
Resolution after handoff (problem confirmed solved, not just closed):
After-hours questions with an owned reply by end of next business day:
Missing items named in "yes" cases (teammate had to ask):
Changes for next week:

Read these measures against your own baseline. Another company's rates do not predict yours.

See a support handoff in a Desk walkthrough: https://www.fullsession.io/book-a-demo?solution=support-desk