Datanovel know-how
How to measure vendor master data quality
Separate completeness, validity and verification before combining them into a readiness measure.
A populated field is not necessarily useful data. A number can have the right structure but belong to another business; a supplier name can look plausible but be out of date. Meaningful measurement distinguishes what was actually checked.
Three different levels of evidence
| Check | What it establishes | What it leaves open |
|---|---|---|
| Completeness | Required information is present. | Whether it is valid or current. |
| Validation | The value meets an applicable format or consistency rule. | Whether it belongs to this supplier. |
| Verification | Evidence from an appropriate source supports a stated fact at a recorded time. | Other facts outside that check, and later changes. |
Make the denominator explicit
Define the population before calculating a percentage. A field may be required for some suppliers and inapplicable to others. Report inapplicable cases separately from missing information; do not penalise every supplier without the same type of registration number or payment identifier.
For each check, show the eligible population, passes, failures and unresolved cases. Keep a failed check separate from a check that could not be completed because the source was unavailable.
Calculate readiness at record level
A record is ready for an agreed purpose when it meets all applicable requirements for that purpose. Calculate this directly from the records. Do not derive overall readiness from the lowest field score. Compute it directly from each record’s applicable requirements.
Illustrative example: in 100 records, 10 have a name issue and a different 10 have a contact issue. Each field scores 90%, but only 80 records pass both. If the same 10 records have both issues, 90 pass. The overlap matters.
Publish the overall result alongside the field-level results. A single percentage cannot explain what the team needs to fix.
Retain the evidence and the limits
Store the rule version, source, check date and outcome with the affected record. A verified VAT response does not establish safe bank details; a reachable mailbox does not prove who controls it. Register information is valuable evidence, but it is not a guarantee of complete or accurate supplier information.
Datanovel implements the agreed checks and reporting in your approved data environment, with exceptions that your team can investigate and resolve.
Official sources for the checks described