A field-service integration is ready when the office can trust its records and recover its exceptions. A demonstration that creates one job is useful evidence, but it is not an acceptance test.
Write the acceptance checklist around the decisions your staff must make. Include customer matching, property selection, service eligibility, category mapping, scheduling authority, follow-up ownership, and the exact words used when an action remains pending.
Require observable evidence for each case
For every scenario, save the fictional input, expected outcome, actual record identifier, action status, and reviewer. Keep sensitive production information out of the test pack. Use approved test accounts or controlled records and document any cleanup.
Validate the API entitlement and supported operations with the vendor's current documentation. For example, Housecall Pro's API overview explicitly cautions that not every feature is available for custom integrations. An acceptance checklist should expose that limitation before deployment.
Use a small but consequential test matrix
Consider a fictional HVAC company launching after-hours intake. Its first release permits routine requests, while dispatchers approve urgent work and uncertain addresses. The checklist should test those boundaries directly.
| Test call | Required result |
|---|---|
| Known caller, known property | Correct existing records selected |
| Same phone, two customers | Clarification without privacy leakage |
| New routine request | Correct owned intake or approved job |
| Out-of-area address | No unsupported service promise |
| Remote timeout | Uncertain state and recovery task |
| Repeated event delivery | No duplicate job |
| Human requested | Tested transfer and fallback |
| Caller changes mind | No abandoned action committed |
Inspect destination fields after each call. Ask the reviewer to explain what the technician or dispatcher will see. A transcript that sounds correct does not prove the service address, category, or appointment window is correct.
Make acceptance conditional on operating readiness
Name the person monitoring failures, the person authorized to change rules, and the person who can reconnect the integration. Require a rollback method and a way to suspend remote writes while preserving intake.
A broad launch provides more automation quickly but exposes more business policy to mistakes. A narrower launch with visible staff review can establish reliable action handling before adding complex rescheduling or commercial workflows.
Record known limitations as part of acceptance, with their fallback behavior. “Not tested” and “unsupported” should remain distinct. If the test environment differs from production, list those differences and schedule a controlled verification after cutover.
Use debugging voice failures to decide which identifiers and logs the office will need during an incident.
Questions about integration acceptance
Who should approve the checklist?
The operational owner should approve business outcomes, and the technical owner should approve integration behavior and recovery evidence.
Must every possible call be tested?
No finite checklist covers every call. Prioritize common paths, consequential mistakes, ambiguous records, and dependency failures, then expand from reviewed exceptions.
What prevents a passing checklist from becoming stale?
Version the accepted configuration and rerun affected scenarios after changes to prompts, fields, categories, permissions, or scheduling policy.