Calculations

How to Calculate Manufacturing Master Data Quality, Cleanup Effort, and Risk Scores

Work through the core master data governance formulas with real inputs, units, and worked numbers, from cleanup hours to FMEA risk scores.

Master data governance runs on four repeatable calculations: cleanup effort in labor hours, completeness as a percentage, an FMEA risk priority number, and a cost roll-up in dollars. Each pulls inputs from a different source system, so the discipline is matching the number to its origin. Cleanup effort comes from a data profiling pass and a timed sample. Completeness comes from a field-fill query against the item or work center table. Risk scores come from a scoring workshop. Cost comes from fully loaded steward rates. Mix the sources and the math is meaningless, so pin each input to its record before you compute anything.

Start with item master cleanup effort, the workload driver behind most cleanup sprints. The formula is base hours equal flagged records divided by records remediated per minute, then required hours equal base hours times one plus the allowance. Take 12,000 flagged records at a measured 12 records per minute: 12,000 divided by 12 equals 1,000 minutes, or 16.7 hours of pure correction. Apply a 25 percent lookup and rework allowance and required effort becomes 20.9 hours. The Item Master Cleanup Effort calculator does exactly this. The trap is the rate: bulk field fills run 10 to 20 per minute, but research records drop to 1 to 2 per minute, a tenfold swing.

Completeness is a straight ratio but you must define the denominator as populated critical fields, not all fields. If 120 work centers each carry 8 mandatory attributes (capacity, calendar, cost center, rate, queue, move, setup, run basis), that is 960 required cells. If a profiling query finds 168 blank or invalid, completeness equals 792 divided by 960, or 82.5 percent. Feed that same population into the Work Center Data Completeness calculator to convert the gap into a remediation budget. Always weight critical fields separately: a blank costing rate breaks standard cost, while a blank description is cosmetic, so a single blended percentage hides the fields that actually stop MRP.

The FMEA risk priority number underpins the ERP Data Quality Score, Routing Accuracy Score, and Unit of Measure Conversion Risk calculators. RPN equals severity times occurrence times detection, each rated 1 to 10, giving a raw range of 1 to 1,000. A recurring blank planning parameter rated severity 6, occurrence 4, detection 3 yields 72. Multiplication, not addition, is deliberate: it forces a defect that is severe, frequent, and hard to catch to dominate the queue. Use one fixed scale across a backlog so scores rank internally, and never average the three factors, which would dilute a genuine severity 9 down to a harmless-looking middle number.

Unit of measure conversion risk deserves its own worked example because the failure is silent. Score the base-to-purchase factor on a part bought by the meter and stocked by the foot. A wrong 1 meter to 3.28 feet factor entered as 3.28 to 1 inverts the order by roughly 10.7x. Rate severity 8 (scrapped run or massive over-order), occurrence 4, detection 7 (no plausibility rule at entry) and the RPN is 224, firmly critical. Run it through the Unit of Measure Conversion Risk calculator. Because detection is usually the cheapest lever, adding an ERP reasonableness check that rejects factors above a threshold drops detection from 7 to 2 and cuts the RPN to 64.

Cost roll-ups follow a variable-plus-fixed structure that keeps quotes defensible. Total equals records times cost each times the true-defect fraction, plus a one-time tooling cost. For duplicates: 800 suspected records at $22 to merge each, 45 percent true positives, plus $6,000 in matching software gives 800 times 22 times 0.45, or $7,920 variable, plus $6,000 fixed, for $13,920 total and $17.40 per record reviewed. The Duplicate Item Cost calculator separates these so finance sees which portion scales. The false-positive rate matters most: loose fuzzy matching at 30 percent true positives wastes review time, while manufacturer-part-number matching pushes true rates above 80 percent.

Governance workload closes the loop by converting recurring transactions into steward hours. Sum monthly creates, changes, reviews, and approvals, divide by processing rate, then apply an allowance for approval routing and rework. If a plant generates 2,400 record actions per month at 6 minutes each, that is 14,400 minutes or 240 hours, times a 1.3 allowance for approval bounce-back, giving 312 hours. At 130 productive steward hours per month, that demands 2.4 full-time stewards. The Master Data Governance Workload and Data Stewardship Capacity calculators handle both sides. Recompute the rate every quarter from actuals, because assumed throughput is the single input that quietly wrecks these estimates.

Tie the chain together before you trust any single number. Completeness and the FMEA scores tell you where records stand and which defects rank first. Cleanup effort and cost roll-ups size the work to close the gap. Workload versus capacity confirms the team can sustain it. Reconcile units at every handoff: hours stay hours, percentages stay per-record, and dollars stay fully loaded. A calculation is only as honest as its slowest input, so validate the remediation rate, the true-defect fraction, and the field-fill query against real samples before you present hours or dollars to a steering committee.

Published 2026-07-01.