An integration report says “last run successful,” while procurement cannot find an order in the other system. The order might be new and not yet due, late, outside the last checked scope, or genuinely absent under a completed comparison. One green indicator cannot distinguish these possibilities.
When buying integration monitoring, first request a report answering which records should have been checked by now and which were actually checked. Differences then become interpretable. Counts from both systems without a shared window or reference can create misleading alerts or conceal real problems.
A report needs an explicit denominator
Imagine a fictional integration sending approved purchase orders to an operational system. An illustrative snapshot contains one hundred source movements: ten excluded because of their status, twenty not yet due under the chosen policy and seventy eligible for comparison. These are teaching figures, not customer results or performance benchmarks.
If only sixty-five of the seventy were checked, the remaining five cannot be labeled missing. Within the checked group, some records may match, some differ and some lack a counterpart at verification time. Display these groups separately with investigable references instead of only an overall percentage.
Four states before the word “missing”
| Suggested label | Report meaning | Required follow-up |
|---|---|---|
| Not yet due | Outside the comparison deadline under a declared policy | Show expected eligibility time |
| Due but unchecked | Completed reads have not covered this record | Assign the coverage gap and its cause |
| Late under review | Past the deadline and requiring rule-based verification | Show age, latest evidence and next check |
| Absent in completed comparison | No counterpart found within documented scope and checks | Create a traceable difference and investigate its cause |
Choose labels suitable for your operations while preserving the distinction between no check and a check result. Even “absent” requires an explanation of matching keys, search method and scope. A reference may have changed or a business rule excluded the record; absence alone does not establish the cause.
What belongs at the top of the report?
Request report extraction time, the latest completed source-reading point, comparison cutoff, timezone and eligible-record definition. Also show target-check completion time. If source collection stops, differences may look small simply because new records never entered the report.
Assign approval of eligibility-window changes and retain their reasons. Expanding the window could reduce apparent overdue counts without improving transfer. Compare successive snapshots knowing that eligible populations may change, rather than assuming raw figures always describe the same group.
Validation is separate work from transfer
AWS DMS documentation distinguishes migrated records from validated records, including pending, suspended and mismatched states. These are product-specific capabilities with support limitations, not guarantees for every integration. They explain why buyers should ask what was compared rather than merely what was sent.
Microsoft's multistep workflow pattern describes durable step states, deadlines and follow-up of stalled work. A successful operational step does not independently prove business-record agreement. Ask the implementer what evidence connects execution status to an understandable comparison outcome, without imposing one architecture on every project.
Also ask about order changes after initial transfer. Finding its reference at the target does not establish arrival of the latest approved revision. Define material comparison fields and how revision or change time appears, so matching counts cannot conceal different content.
Who can close a difference?
Give every difference a source reference, target reference where available, discrepancy type, latest check, follow-up owner and next action. Preserve closure reasons and evidence: verified late arrival, approved exclusion or repair followed by another check. Removing a row from the screen is not an explanation.
For recovery behavior, review retry acceptance; for transfer timing, review synchronization frequency selection. When requesting integration implementation from Nextwave, attach a coverage-report example to purchase interpretable operational follow-up.
Sources for state distinctions
Documentation on AWS DMS data validation and the Scheduler Agent Supervisor pattern was opened on 4 November 2026. The example and table are proposals without actual integration operation or measurement. AI assisted preparation, with information and translation review.
