Quality & InspectionFree calculatorPDF, Excel, Word export

Data Quality Sigma Level Calculator

Manufacturing teams already understand sigma levels and DPMO, so expressing data quality the same way makes ERP data problems legible to operations and quality leaders. Enter your audit counts and get an error rate, DPMO, sigma level and the share of records with every checked field right.

Your numbers

Records inspected

records

Fields checked per record

fields

Each checked field is one opportunity for a defect.

Defective fields found

defects

Sigma shift convention

Your results

Defects per million opportunities

12,000

Sigma level

3.7571

Total opportunities

70,000

Field error rate

1.2%

Records expected to be fully correct

84.4%

Rolled yield assuming defects are independent across fields.

Sigma conversion assumes the standard normal approximation with the stated shift. Defects that cluster inside the same records will make the real clean-record share higher than shown.

Save results:

Sigma level vs defective fields found

Sweeps Defective fields found from 10 to 3000, holding your other inputs fixed.

Continuous data quality rules

Optionally ask how your audit rules could run on a schedule with trend reporting.

Optional. The tool and every export work without this. No spam.

How to use it

Audit a random sample of records, using the sample size calculator if you need a justified count. Decide which fields are in scope, such as unit of measure, lead time, planner code, commodity code, tax class and lot control flag, and count each field in each record as one opportunity. Enter the records inspected, fields per record and the number of defective fields found. Choose the sigma shift: most Six Sigma conventions quote long-term sigma with a 1.5 shift, so 3.4 DPMO equals 6 sigma. Use the same fields and rules every time to make comparisons meaningful.

The method

Opportunities equal records times fields. DPMO equals defects divided by opportunities times one million. The sigma level is the inverse of the standard normal distribution at one minus the defect rate, plus the shift. Record-level yield is one minus the defect rate, raised to the number of fields, which shows how a seemingly modest field error rate wrecks whole-record accuracy. For example, 99 percent field accuracy over 14 fields gives about 87 percent fully correct records. The chart shows sigma falling as defects rise, and how slowly sigma improves at the high end, where each fraction costs far more effort.

DPMO = defects / (records x fields) x 1,000,000
Sigma = NORMSINV(1 - defect rate) + 1.5
Clean record share = (1 - defect rate)^fields

Interpretation and cautions

Do not chase a six sigma number for its own sake. Weight fields by consequence: a wrong planner code is annoying, a wrong lot control flag or tax class can stop shipments or break compliance. Track critical fields separately with tighter targets, and report the sigma level alongside the clean record share, since people grasp the latter immediately. Common mistakes are changing the field list between audits, mixing completeness defects with accuracy defects without labelling them, and auditing only records that were recently touched, which are often cleaner than the long tail of dormant records.

Frequently Asked Questions

What sigma level is typical for ERP master data?

Many organizations that measure it find field-level accuracy between about 95 and 99.5 percent, which corresponds to roughly three to four sigma with the 1.5 shift convention. Your own audit matters far more than any benchmark.

Should I count missing values as defects?

Yes, if the field is mandatory for your processes. Many teams track completeness and accuracy separately, since the fixes differ: completeness needs required fields and defaults, accuracy needs validation and verification sources.

Why is the clean record share so much lower than field accuracy?

Each record has many fields, and any single defect makes the record imperfect. Raising the field accuracy to the power of the field count compounds small errors, which is why record-level targets feel harsh.

Is the 1.5 sigma shift appropriate here?

It is a convention from process capability work, not a law of data. Use it if your organization reports Six Sigma metrics that way, otherwise choose no shift and say so on the report.

Once your rules are defined, they can run continuously instead of by annual sample. Netray can help set up read-only rule checks across your ERP tables and a dashboard of defect trends.

Talk to an Expert Explore More Free Tools

Related Tools

Quality & Inspection

More Quality & Inspection tools

See all

Back to All Tools