A payment problem in a fintech app integration can involve several stages: the original request, a provider response, a receiving record and the business’s later review. A Great Southern business needs to understand those stages before deciding what a failed-looking result means.
This checklist helps prepare questions and responsibilities. It does not diagnose a live transaction or instruct a business to create a failure, repeat a payment or change an accounting record.
1. Identify the actual connected route
Name the relevant source app, payment service, receiving system and business contribution. A connected logo does not explain which party handles each stage.
Keep technical delivery, payment outcome and record preparation distinguishable. A successful transfer of information to a bookkeeping view is not automatically proof that every underlying financial question is resolved.
Check actual service and plan conditions rather than assuming all integrations share the same scope.
2. Record the original request’s identity
Identify the relevant reference, source and intended obligation. Similar amounts can relate to different transactions, while different labels can concern one original request.
Keep the relationship understandable for the next authorised reviewer. Do not copy an unnecessary financial archive or expose credentials in general incident notes.
Record what is known directly and what remains a provisional interpretation.
Retain the sequence of relevant observations. A later response or corrected source should not make an earlier reviewer appear to have known something that was unavailable when the initial decision was made.
3. Separate confirmed failure from an unknown result
A provider-confirmed failure is different from an unclear response. A stalled screen does not establish that the original payment never occurred.
NAB Gateway’s timeout information illustrates why an uncertain response needs investigation of the original outcome. It is not a universal retry interval or reversal instruction.
Use the relevant service’s supported checking route before another consequential request.
4. Check the meaning of a delivery status
An integration can report on delivery to one receiving stage without describing every downstream action. Clarify what the actual status confirms and what it leaves unknown.
As a bounded example, GitHub’s webhook guidance states that failed deliveries are not automatically redelivered. Do not generalise that behaviour to all services or treat a delivery failure as universal proof of non-execution.
5. Understand the limits of duplicate protection
Ask which supported mechanism identifies a repeated request and under what conditions. An internal duplicate note is different from a provider’s technical control.
Adyen’s idempotency documentation describes key, scope and handling conditions. That example does not establish that every fintech app safely repeats every payment.
No API call or real repeat transaction is needed to ask what the proposed integration supports.
- Map the actual route and original reference.
- Separate payment, delivery and record meanings.
- Investigate uncertain results before repetition.
- Check schedules, review duties and actual roles.
- Record remaining questions and their owners.
6. Ask separately about scheduled activity
A particular failed attempt does not explain the position of every later scheduled item. Identify what the actual service says about the relevant arrangement.
For example, NAB Portal Pay troubleshooting distinguishes a failed payment from later scheduled recurring attempts. Do not infer that the underlying obligation disappears or that all providers behave identically.
7. Retain the financial review responsibility
Information entering another system does not complete every check. The receiving reviewer still needs the relevant account, period, source and explanation of differences.
MYOB’s reconciliation guidance explains that accounts still need reconciliation with bank feeds. Do not invent an adjustment, accounting classification or tax treatment to make an unexplained difference disappear.
8. Consider actual operating conditions
A Great Southern workplace can have different dependencies from another site using the same app. Consider actual device, connection and supported offline behaviour rather than blaming every uncertainty on the region.
ACMA’s coverage map information explains predictive outdoor limits. A map is not a guarantee of service inside a particular building or vehicle, and a visible screen is not proof that the request arrived.
9. Assign investigation and decision roles
Separate gathering evidence, obtaining a technical explanation and approving a financial commitment. Actual service permissions and internal business authority may differ.
Identify appropriate cover and genuine contact routes. A support enquiry acknowledged is not the same as a transaction outcome established.
If several parties are involved, preserve the precise question sent to each. An answer about information delivery should not silently become assurance about settlement, accounting completeness or a different connected service.
10. Review the plan through bounded examples
Use suitable fictional material in actual supported conditions to examine a confirmed failure, missing response and amended record. Keep live money, customer messages and forced interruptions outside the exercise.
Record assistance, untested stages and the next owner. For Great Southern businesses, a useful integration failure plan preserves uncertainty until evidence clarifies it and keeps consequential authority separate from technical convenience.