Business software integration moves information or instructions between systems. For an Albany business, useful measures should show whether the receiving workflow gets what it needs, rather than simply celebrating a larger number of automated events.

These problems and responses help define measures that reflect real work. They apply to the actual integration being examined; a product’s documented mechanism should not be assumed to exist in every connected service.

Problem: the integration has no defined business finish point

A connection can be technically active while leaving the receiving colleague uncertain about what to do. Counting transmitted records does not establish that the intended work is complete.

Define the business event first. For an enquiry-to-job workflow, that might be an understandable request reaching the appropriate review queue with the information needed for the next decision. Keep that event distinct from a connection test or a message sent by the source system.

Problem: every technical event is counted as new work

One business item can generate several updates or attempts. A rising event total may therefore reflect repeated activity around the same item rather than additional completed work.

Ask how the relevant systems identify the business item and its events. Keep the original reference and meaningful distinctions between creation, update and repeated delivery. The measure should explain which event is counted and why. A total without that definition can be precise yet uninformative.

Problem: duplicate protection is assumed rather than documented

Repeating an instruction can have different consequences depending on the particular service and mechanism. A general assurance that automation handles retries does not explain the actual conditions.

Adyen’s idempotency documentation describes a conditional mechanism for supported API requests, including key scope and error handling. That named example shows why repeated attempts and unique operations require careful distinction. It does not establish a universal guarantee for every integration, and an uncertain consequential outcome should be verified before repetition is considered.

Problem: received is treated as understood and accepted

A receiving system may acknowledge a message before the business can use its content. A record may arrive with an unknown category, missing context or an unresolved validation question.

Separate receipt, usable content and completion of the intended receiving task. Ask what each status actually means in the particular product. If the workflow requires human interpretation, record that stage rather than describing the technical acknowledgement as the final business result.

Problem: an exception vanishes from the success total

A measure can look favourable if it counts only items that followed the ordinary route. Missing information or unresolved items may be left outside the report without explanation.

Define the population being reviewed and account for the relevant exceptions. Record whether an item is awaiting clarification, excluded for a stated reason or outside the chosen period. An uncounted item should not silently become evidence that no problem occurred. Keep the question open until the appropriate source explains it.

Problem: elapsed time is measured across different boundaries

One system’s creation time may not represent the moment the customer began the task. A receiving timestamp may not establish when a colleague had enough information to proceed.

Choose a start and finish related to the actual work, and document the meaning of the timestamps used. Separate waiting for missing information from processing or interpretation where that distinction matters. Do not compare unlike periods as though a shorter interval necessarily proves an improved whole workflow.

Problem: logs are treated as complete business evidence

Technical records can help explain system activity, but their purpose and coverage need to be understood. A log entry may describe an attempt or configuration change rather than the underlying commercial outcome.

ASD’s event-logging guidance focuses on useful activity details and protected, reviewed logs. Use appropriate technical evidence alongside the records needed for the business task. Do not infer payment completion or financial authority from a generic system success label.

Problem: a measure changes meaning during the review

A revised field, category or workflow can make an old total incomparable with a new one. The apparent improvement may arise from changed counting rules rather than better work.

Retain the definition, source and period of each measure. Business.gov.au’s performance guidance recommends linking metrics to goals and checking platform documentation. Explain a material definition change before drawing a trend across it.

Problem: the review cannot produce a practical response

Measures should help decide what to clarify or improve. Use a bounded invented example and supported non-live conditions where examination might otherwise affect records, messages or account actions.

  • The business item and intended receiving result.
  • The technical stages and their actual meanings.
  • The exceptions and missing evidence.
  • The next question, source and responsible person.

For an Albany business, an integration measure becomes useful when it explains whether real work can proceed. Activity counts support that assessment only when their definitions and limitations remain visible.