Quality assurance dashboard: defects, rework, and trends

A quality assurance dashboard compares inspection failures, rework, scrap, and unresolved quality exceptions from the same inspected population. Its most useful check is simple: every failed unit must have a recorded disposition or remain visibly open. Counts show the workload; rates make sites with different inspection volumes comparable.

Start with the inspection population

Use the initial inspection register as the starting file. Keep a unique unit identifier, site, product family, inspection timestamp, first-pass result, and defect code. A separate disposition export can supply rework completion, scrap, or hold status, linked back to that same identifier.

The distinction between a defective unit and a defect matters. NIST distinguishes nonconforming items from individual nonconformities: a unit can have more than one recorded problem. Counting every defect-code row as a failed unit would inflate the failure rate.

For this dashboard, a failed unit is a unique unit that did not pass its initial inspection. Count that identifier once in the failure numerator. Use defect events in a separate breakdown when investigating what went wrong.

First-pass failure rate = Unique units failing initial inspection / Unique units initially inspected

First-pass yield = Units passing initial inspection / Unique units initially inspected

NIST defines the proportion nonconforming as nonconforming items divided by the relevant item population. Here, that population is the inspected cohort, rather than all production at the plant.

A worked example across two US sites

Consider an illustrative US manufacturer of metal housings. Its Ohio and Indiana plants inspect completed units during the week ending October 16, 2026. The dashboard freezes disposition status at the weekly review. All successful rework shown below has passed reinspection; units still awaiting a result remain on hold.

First-pass successes are inspected units minus failures. The calculation and the resulting rates are:

Percentages are rounded to two decimal places after calculation. Rework completion share uses the same inspected-unit denominator: Ohio is 120 / 4,000, Indiana is 80 / 2,000, and the combined result is 200 / 6,000.

Indiana has fewer failures in absolute terms, but a higher failure rate. That makes it the priority for investigating first-pass performance. Ohio still has the larger completed rework volume, so it may need more rework capacity even while its failure rate is lower.

For the workbook implementation of these measures, use the separate spreadsheet quality dashboard guide.

Prove that the units reconcile

The first check closes each site's failed population:

The combined accepted population at the cutoff is 5,700 first-pass successes plus 200 successful reworks, or 5,900 units. Its share of inspected units is 5,900 / 6,000 = 98.33%. Label it accepted by cutoff, because the 20 held units have no final result yet.

The broader reconciliation is 6,000 = 5,900 accepted + 80 scrapped + 20 held. A held unit cannot simultaneously appear in accepted output. If a disposition export contains successive status updates, select the latest valid status at the cutoff rather than summing the history.

Errors in identifiers and joins need their own data-quality dashboard; they should not appear as manufacturing defects.

Add cost without hiding the operational result

The same company records rework labor at a fully loaded hourly rate supplied by finance. Scrap cost below is the recorded carrying cost written off, with no recovery proceeds in this example.

Tracked failure cost is rework labor plus scrap cost: 3,850 + 640 = $4,490. This is a defined subtotal. Inspection overhead, customer returns, freight, and other costs are outside the example. See the separate quality-cost dashboard example for a wider cost review.

Now change one input: increase the common labor rate from $35 to $42 per hour. Keep hours, failures, rework completions, scrap, and holds unchanged. Rework labor becomes 110 × 42 = $4,620; tracked failure cost becomes 4,620 + 640 = $5,260. The increase is 110 × (42 − 35) = $770.

The failure rate remains 5.00%. The cost increase comes entirely from the labor rate, so the dashboard should not describe it as deteriorating inspection performance.

Turn the review into an action list

Let the quality manager own inspection definitions and exception disposition. Site leaders own the investigation of recurring failures. Finance owns labor rates and scrap valuation. Keep each hold linked to its unit or lot, current reason, accountable role, and next review date.

For a trend, repeat the same cohort calculation for each closed week. Show the inspected count beside the failure rate and retain the cutoff convention. A recent cohort with more unresolved holds is less mature than an older one, even when its first-pass result is already complete.

Where it goes wrong

• Counting defect codes as units. Several defect rows can refer to the same failed housing. Deduplicate unit identifiers before computing a unit failure rate.
• Adding reinspection to the denominator. Repeat tests do not create newly inspected units in a first-pass measure.
• Combining unmatched cohorts. This week's completed rework may belong to earlier inspections. Link dispositions back to the original cohort.
• Averaging site percentages. Divide combined failures by combined inspected units. Small sites should not carry the same weight as larger sites.
• Calling a subtotal total quality cost. State which labor and scrap accounts are included, and reconcile them without adding the same charge twice.

Review your own inspection file in Covirage

Bring an inspection export and the matching disposition and cost records to Covirage. Covirage maps the columns; its tools compute and check the figures, and the external AI model explains the results with every figure traced to its rows. Create a free Covirage account, then set up your business and bring a sample of your own data.

See Covirage for industrial manufacturers for the manufacturing context.