The report looks complete until someone tries to contact a client using an old email address. Operational data-quality checks should test whether information is fit for the decision or task it supports. Start with important fields, a defined rule and an owner who can resolve failures.
Photo by Rodrigo Rodrigues, WOLF ART on Unsplash.
The UK Government Data Quality Framework distinguishes completeness, uniqueness, consistency, timeliness, validity and accuracy. These are useful prompts for business records too, provided the checks match your actual use rather than becoming a generic audit.
Choose the consequence before the check
Identify a record whose failure disrupts work, such as a delivery address, approval contact or promised date. Explain what goes wrong when it is missing or incorrect. Use that consequence to decide how quickly a failure needs attention.
Do not treat every empty field as equally important. An optional note and a missing dispatch address need different handling.
Build a small quality-check register
Field or record: the information being checked.
Intended use: the decision or task depending on it.
Rule: what a passing record looks like.
Method: automated validation, reconciliation or manual sampling.
Frequency: when the check is useful in the workflow.
Owner and response: who investigates and what happens meanwhile.
Evidence: failure, correction and possible source problem.
A valid format does not prove a value is accurate. A plausible date can still be the wrong date. Where accuracy matters, compare against a suitable authoritative source or seek confirmation through the approved process.
An illustrative client-record check
A team preparing delivery appointments checks that every booking has a current client contact and confirmed location. Missing information creates a visible exception before routing begins. The record owner resolves it rather than the scheduler guessing.
The team also reviews repeated failures by source. If one form regularly produces incomplete locations, fixing that form is more useful than repeatedly cleaning the output.
Keep the response proportionate
Some failures can wait for routine correction. Others should stop the affected task until reviewed. Agree those boundaries with the service owner, and avoid allowing an automated rule to overwrite uncertain information without suitable control.
Related guidance: practical process-mapping guide.
If unreliable records keep interrupting delivery, Hili Consulting can help select practical checks and clarify responsibility for preventing the same errors returning.
