01
Routine inquiry
Make the call
Ask about a service you actually offer in your service area.
Check the outcome
The request is captured accurately and reaches the intended next step.
A natural conversation is only one part of a successful call. The appointment, callback request, or handoff needs to match what the caller was told.
Set the rules before making a call
Write down your service area, supported job types, opening hours, approved booking rules, and callback owner. Decide what the assistant should do when information or availability is missing. Define the fallback when a person or connected system cannot be reached.
Use fictional customer details and a designated test calendar or queue. Agree on the expected outcome before each call. Record the call time and resulting record so you can compare the conversation with the actual outcome.
This is a suggested evaluation checklist, not measured customer results or a guarantee of a product’s capabilities. Use the checks that match your configured service.
THE CALL CHECKLIST
Make these ten calls.
Test one situation at a time. Compare the expected outcome with what actually happened.
02
Outside your service area
Make the call
Give a location your business does not serve.
Check the outcome
The answer follows your boundary and does not promise an appointment your team cannot fulfill.
03
Unsupported work
Make the call
Ask for a job your business does not perform.
Check the outcome
The assistant explains the limitation without inventing a service or referral.
04
Unknown budget
Make the call
Request an estimate but say you do not know your budget.
Check the outcome
The outcome follows your qualification policy. Missing budget information is recorded accurately.
05
Unavailable appointment
Make the call
Ask for a slot you have deliberately made unavailable.
Check the outcome
The assistant does not confirm that slot. Any alternative or callback request matches your approved process.
06
After closing
Make the call
Call just after the configured closing time.
Check the outcome
The correct after-hours path runs using the business's timezone. Callback wording matches the team's actual coverage.
07
Human unavailable
Make the call
Ask for a person while the designated test recipient is unavailable.
Check the outcome
The fallback captures the request and routes it to an accountable owner without claiming a transfer succeeded.
08
Corrected details
Make the call
Give a fictional address or callback number, then correct it.
Check the outcome
The final record contains the correction and the assistant confirms the essential details.
09
Repeat caller
Make the call
Call again about the same fictional request.
Check the outcome
The second interaction follows your repeat-contact policy. Check for conflicting bookings, records, or follow-up tasks.
10
Connected-system failure
Make the call
In an isolated test setup, simulate an unavailable calendar or CRM if that connection is in scope.
Check the outcome
The caller hears the approved fallback. No booking or record update is claimed unless it succeeded.
Check beyond the conversation.
For each test, inspect the destination your team actually uses. If the assistant says an appointment is confirmed, verify it in the calendar. If it promises a callback, verify the request has an owner and an agreed response window. Where a CRM connection is configured, compare the saved fields with the caller's final details.
An internal alert does not establish that a customer was contacted. Track attempted callbacks and completed conversations separately.
Record a decision for each scenario.
Use Pass, Needs work, or Not applicable. A pass means the observed conversation and downstream outcome match the expectation you wrote before the call. An untested scenario stays Not tested.
Record what failed, who will fix it, and when it will be retested. As a suggested launch gate, resolve false booking confirmations, lost callback requests, and failed handoffs before routing real callers into the affected path. Repeat the relevant checks after changing prompts, hours, routing, or integrations.
YOUR WORKING COPY
Keep every test accountable.
A printable result sheet for all 10 calls. Record the expectation, observed outcome, status, fix owner, and retest date.
Testing with Lacy.ai
Lacy's testing guide explains how to test incoming calls by calling the number assigned to an assistant, and outgoing calls through the assistant's test-call control. Its automation documentation describes post-call actions, including contact updates and follow-up workflows. Verify the actions configured for your deployment; documentation alone does not establish that your account has a working connection.
Keep the result sheet alongside your call-flow requirements so everyone evaluating the setup uses the same expectations.
LACY.AI · HOME SERVICE PLAYBOOK
AI receptionist test result sheet
Business: ____________________ Tester: ____________________
Date / timezone: ____________________ Configuration version: ____________________
Write the expected outcome before each call. Record Pass, Needs work, Not applicable, or Not tested. Keep call references free of unnecessary personal information.
| Test | Expected outcome | Observed outcome / call reference | Status | Fix owner / retest date |
|---|---|---|---|---|
| 1 | ||||
| 2 | ||||
| 3 | ||||
| 4 | ||||
| 5 | ||||
| 6 | ||||
| 7 | ||||
| 8 | ||||
| 9 | ||||
| 10 |
Release decision and remaining issues: ___________________________________________