How do you calculate a production data quality score?

Multiply three 1-10 ratings: severity of the data defect's impact, how often it occurs, and how hard it is to detect. With severity 7, occurrence 5, and detection 6 the raw product is 210, scaled here to a 6.05 priority number.

What is a good data quality risk score?

Lower is better. There is no universal cutoff, but most teams set an action threshold (commonly any item above a chosen percentile of their backlog) and treat high-severity items with poor detection as must-fix regardless of frequency.

Why use severity x occurrence x detection for data quality?

It borrows the proven FMEA logic so a rare but catastrophic and undetectable defect (e.g. silent genealogy gaps) ranks above a frequent but obvious one that operators catch immediately.

How should I rate detection difficulty for MES data?

Rate 1 if automated validation or a dashboard flags the defect instantly, and 10 if it only surfaces during a downstream audit or recall. A 6 means it usually escapes routine checks but is found before it reaches customers.

RPN vs simple error rate for data quality?

An error rate tells you how often data is wrong; the RPN weights that by how damaging and how invisible each error is, so it prioritizes fixes better than frequency alone.

How often should we rescore data quality risk?

Re-rate after any control change or new integration, and review the full backlog quarterly so occurrence and detection ratings reflect the controls you have actually deployed.

MES, MOM & Shop-Floor Data Systems calculator

Production Data Quality Score Calculator

The Production Data Quality Score adapts the FMEA risk priority number to MES and shop-floor data integrity, rating each data defect on severity, occurrence, and detection. Data engineers, MES administrators, and quality teams use it to decide which broken feeds, mistyped operator entries, or missing scan events deserve controls first. On a connected line, a single corrupt genealogy record or a drifting sensor tag can silently poison OEE dashboards, traceability, and SPC. Scoring the risk turns a vague 'our data is messy' complaint into a ranked, defensible backlog.

What this calculator does

Score the risk of poor production data quality using an FMEA-style approach: severity of impact on decisions, frequency of bad data occurrence, and difficulty of detection.
Use during data governance reviews to prioritize which data quality issues to fix first. Higher scores indicate data problems that cause the most damage and are hardest to catch.
It multiplies severity, occurrence, and detection ratings (each 1-10) into a single data quality risk priority number that ranks competing data defects.

Formula used

Data quality risk score = severity x occurrence x detection
Higher scores indicate data issues that need immediate attention and controls.

Inputs explained

Severity of data quality impact on downstream decisions:
How often this data defect occurs:
How hard the defect is to detect before it propagates:

How to use the result

Use it during MES data audits, after a bad-data incident, or when building a data governance backlog and you need to triage many issues with limited engineering time.
The score is ordinal, not absolute — a 200 is not literally twice as risky as a 100, and ratings depend on consistent team calibration, so compare scores only within the same rating rubric.

Common questions

How do you calculate a production data quality score? Multiply three 1-10 ratings: severity of the data defect's impact, how often it occurs, and how hard it is to detect. With severity 7, occurrence 5, and detection 6 the raw product is 210, scaled here to a 6.05 priority number.
What is a good data quality risk score? Lower is better. There is no universal cutoff, but most teams set an action threshold (commonly any item above a chosen percentile of their backlog) and treat high-severity items with poor detection as must-fix regardless of frequency.
Why use severity x occurrence x detection for data quality? It borrows the proven FMEA logic so a rare but catastrophic and undetectable defect (e.g. silent genealogy gaps) ranks above a frequent but obvious one that operators catch immediately.
How should I rate detection difficulty for MES data? Rate 1 if automated validation or a dashboard flags the defect instantly, and 10 if it only surfaces during a downstream audit or recall. A 6 means it usually escapes routine checks but is found before it reaches customers.
RPN vs simple error rate for data quality? An error rate tells you how often data is wrong; the RPN weights that by how damaging and how invisible each error is, so it prioritizes fixes better than frequency alone.

Last reviewed 2026-05-12.