Why an Integration Can Succeed While Records Are Missing
A successful integration run means the workflow reached its definition of success. It does not necessarily mean every source record is present and correct in the target.
Execution success and data success are different
An integration engine primarily observes execution: requests, steps, responses, retries, and workflow completion. Data correctness is a separate concern.
If the workflow completes without triggering a fatal error, the run may be marked successful even when the resulting dataset is incomplete.
How records can disappear
- A source query excluded a record unexpectedly
- Pagination stopped before every record was retrieved
- A downstream API accepted only part of a batch
- Conditional workflow logic skipped a record
- A retry or partial failure was not reflected in the final execution status
Why logs are not always enough
Logs explain what the workflow attempted to do. Reconciliation verifies the state that exists after those attempts are finished.
The safest way to answer whether every record arrived is to compare the actual source and target data.
A simple verification model
After the integration completes, run a source-to-target comparison using a stable record ID and the business-critical fields. Any source record without a matching target record should be surfaced as Missing.
Do not stop at integration success.
Compare the actual source and target data after the workflow finishes.
Start verifying→