Search for a command to run...
Articles tagged with #data-quality
Moving customer deduplication off an on-prem monolith and into a Bronze/Silver/Gold lakehouse sounds like a straight port. It is not. There is no direct JOIN to lean on, the old "keep the oldest ID" cleanup quietly nulls out half your accou
When you ingest financial data from dozens of sources, the spec and the actual files disagree more often than they agree. One source's "balance" is the total, another's is only what is current; one reports aging as a percentage, another as
On a retail project last year, we spent an afternoon chasing down Opportunity records with negative amounts that kept slipping past a validation rule we'd already written and tested. The culprit turned out to be a nightly sync job inserting
Row level checks catch nulls and bad types, but they wave through data that is well formed yet wrong, like a contributor whose arrears drop 90 percent in a month. The fix is a Hybrid Baseline check that validates each batch against its own
The same person can land in your data as three records, each right about one field and stale about another. Survivorship is the rule that picks one golden record per customer. Here is how the customer type hierarchy decides the winner, and
Banking client's morning file arrived with 50 rows instead of 10,000. Automation sent reminders to those 50 and skipped the other 9,950. Verification Activity is the check that would have stopped it.
