← Back to Field Notes

September 24, 2025 · The Bloomfield Team

What Medical Device Shops Get Wrong About Quality Data

FDA 21 CFR 820 requires medical device manufacturers to maintain quality records. ISO 13485 requires it. Every customer audit checks for it. The result: medical device contract manufacturers collect an enormous volume of quality data. Inspection reports, non-conformance records, CAPA documentation, device history records, process validation protocols. Binders full of it. Servers full of it.

Almost none of it gets used to actually improve operations.

The Compliance Trap

The fundamental problem is that quality data in medical device manufacturing gets collected for regulatory compliance rather than operational improvement. The systems, the formats, and the workflows are all designed to satisfy an auditor's checklist. They are not designed to answer the questions that would make the operation better: which processes produce the most variation, which suppliers deliver material that causes the most rework, which operators consistently hold tighter tolerances, and which product families carry the highest cost of quality.

The data to answer every one of those questions already exists inside the quality system. It sits in inspection records, NCR databases, and CAPA files. But it sits in a format that was designed for retrieval during an audit, organized by document number and date, rather than by the operational dimensions that would make it useful for improvement.

For a broader look at how data accessibility affects manufacturing operations, see our guide to manufacturing knowledge management.

Three Things Most Shops Get Wrong

1. Recording Data Without Analyzing Trends

A medical device shop runs 200 first-article inspections per quarter. Each one generates a detailed measurement report documenting every critical dimension against specification. Those reports get filed. They satisfy the DHR requirement. What they do not do is feed into a trend analysis that would show the quality manager that bore diameter measurements on parts machined on cell 3 have been drifting toward the upper control limit for the past eight weeks.

The inspection data exists. The trend analysis does not happen because the data lives in individual PDF reports rather than in a structured database that supports time-series queries. That is a system design problem, and solving it does not require replacing your quality system. It requires connecting the data it already collects to a tool designed for analysis.

2. Treating CAPAs as Events Rather Than Data

A corrective action gets opened, investigated, resolved, and closed. The CAPA record documents what happened, what the root cause was, and what corrective action was taken. That record then goes into a folder and stays there until someone opens a similar CAPA and nobody remembers the last one existed.

CAPAs are pattern data. Over three years, a medical device shop might generate 40 to 80 CAPAs. Those records, in aggregate, reveal which process steps generate the most failures, which suppliers are implicated most often, and which corrective actions actually prevented recurrence versus which ones were paperwork exercises. That pattern analysis drives systemic improvement. Individual CAPA resolution does not.

3. Disconnecting Quality Data from Quoting

When a medical device shop quotes a new job, the estimator builds the price based on material, cycle time, setup time, and secondary operations. What rarely factors into the quote is the historical cost of quality for similar parts: the inspection time, the documentation overhead, the rework rate, and the CAPA probability.

On a complex medical device component with tight tolerances and full dimensional inspection requirements, the cost of quality can represent 8 to 15% of total job cost. If the estimator does not account for it, the shop wins the job at a price that does not cover the true cost of producing it to spec. Connecting quality data to the quoting process means the gap between quoted and actual costs closes, and margins on medical device work become sustainable rather than erratic.

What the Fix Looks Like

The fix is straightforward conceptually and requires discipline to execute. Three changes matter most.

First, structure the quality data for analysis. Inspection measurements need to feed into a database that supports filtering by part, by machine, by operator, by time period, and by dimension. NCRs need to be categorized by root cause type, not just documented as individual events. This does not require new software. It requires exporting data from your current systems into a format designed for retrieval.

Second, run monthly quality trend reviews that go beyond the regulatory requirements. Look at Cpk trends by process and by machine. Look at CAPA frequency by product family. Look at cost of quality as a percentage of revenue by customer and by part type. These reviews produce the actionable findings that drive improvement.

Third, feed quality cost data into the quoting process. When the estimator pulls up a comparable past job, they should see the inspection hours, the rework events, and the documentation overhead alongside the material and machining costs. That complete picture produces quotes that sustain margins on regulated work.

Medical device manufacturing demands the highest standards of quality documentation in the industry. The shops that turn that documentation from a compliance cost into an operational advantage will outperform their competitors on margins, on delivery, and on the audit results that keep them qualified for the work. The data is already there. The systems to use it are buildable now.