The Cheapest Place to Catch a Defect

The cost of a defect multiplies roughly by ten at every stage it survives. A bad resin lot caught on the receiving dock costs a rejection notice and a replacement shipment. The same lot caught after molding costs 8,000 scrapped parts. Caught in the field it costs a recall, and a consumer recall of 20,000 units routinely runs past $400,000 once logistics, replacement, and support are counted.

Incoming quality control is the gate that stops that multiplication. It is also the quality function most small hardware companies skip, because it feels like inspecting things you already paid someone else to get right.

What Belongs on the IQC Checklist

IQC is four separate checks, applied with different intensity depending on the material and how much damage it can do.

  • Identity. Is this the part you ordered? Manufacturer part number, revision, resin grade including the exact suffix, color code, temper, alloy. A substitution from PC/ABS to plain ABS, or 6061-T6 to 6063, is invisible on the dock and catastrophic in the field.
  • Quantity and packaging. Piece count, moisture barrier bags intact, desiccant present and unspent, ESD packaging where required, no crush or water damage.
  • Documentation. Certificate of conformance, certificate of analysis, mill test report, RoHS and REACH declarations, food-contact letters, lot and date codes traceable to the manufacturer.
  • Physical and functional verification. Actual measurement against your drawing: critical dimensions, hardness, plating thickness, electrical parameters, melt flow index for resin, torque for fasteners.

The fourth item gets dropped most often and is the only one that catches a real problem. Anchor it to a signed reference part, using the same logic as the golden sample you hold for finished goods.

A Certificate of Analysis Is Not a Test

A CoA is a supplier's statement about a lot. A certificate of conformance is weaker still, an assertion that the goods meet spec without any data attached. Both are useful and neither is verification.

The failure modes are mundane rather than sinister. The CoA describes a lot that was blended with another before shipment. The values are copied from the previous lot's report. The test was run at the resin producer, and the material then passed through two distributors and a repackager. Or the certificate is genuine and simply does not cover the property you care about.

Treat documentation as necessary but not sufficient. Verify physically on anything whose failure would reach your customer, and reserve document-only acceptance for low-risk, high-history parts. In electronics this is not optional, because the entire counterfeit trade depends on paperwork looking correct, a problem detailed in counterfeit electronic components.

Sizing a Sampling Plan

You cannot test everything and do not need to. A standard sampling plan such as ANSI/ASQ Z1.4 gives a defensible sample size for a given lot size and acceptance quality limit. For a lot of 5,000 pieces at general inspection level II with an AQL of 1.0, the plan calls for a sample of 200 with acceptance at 5 defects. The same table logic drives finished-goods AQL inspection.

Layer risk on top of the table. Classify every purchased item as critical, major, or minor. Critical items, meaning anything affecting safety, regulatory compliance, or a function that cannot be reworked, get tightened inspection or 100% screening. Minor items such as packaging inserts get skip-lot inspection after a clean history. Then let performance move the level: twelve consecutive accepted lots earns reduced inspection, one rejection moves a supplier to tightened until they earn their way back. Publishing that rule changes supplier behavior more than any complaint email.

Separately, run a full first article inspection on the first shipment from any new supplier, tool, or revision. FAI verifies every drawing characteristic once; IQC is the ongoing sample. Confusing the two leaves a gap exactly where new-supplier risk is highest.

Supplier Scorecards and Lot Quarantine

IQC data is worthless if it lives in a binder. Two mechanisms turn it into leverage.

A supplier scorecard tracks, per supplier per quarter, lot acceptance rate, defects per million, on-time delivery, documentation completeness, and response time on a corrective action request. Share it with the supplier every quarter. Suppliers manage what they are measured on, and a factory that sees its own numbers next to a stated threshold will fix problems you never had to escalate. This is the same discipline that keeps a young hardware supply chain from degrading silently as volumes grow.

A quarantine and lot isolation procedure decides what happens to a rejected lot. Segregate it physically in a marked area, never on the production shelf, and tag it with a numbered nonconforming material report. Record the decision as one of four outcomes: return to supplier, sort at supplier cost, use as-is under a documented deviation, or scrap. Keep lot traceability from incoming lot number through to finished goods serial numbers, because when a problem escapes you want to recall 900 units rather than everything you have shipped.

Deviations deserve a rule of their own, because every company grants them under schedule pressure. Require a named approver, an expiry, and a written rationale, or "use as-is" quietly becomes the standard and your spec becomes fiction.

Running IQC Without a Quality Department

A three-person hardware company can run credible IQC with a few hundred dollars of equipment and a disciplined hour per shipment.

Buy the basics: digital calipers, a micrometer, a bench scale accurate to 0.01 g, a pin gauge set, a durometer if you use elastomers, and a coating thickness gauge if you specify plating. Write a one-page inspection instruction per part number listing the three to five characteristics that matter, the sample size, the acceptance criterion, and a photo of a good part. Record results in a shared spreadsheet with lot numbers, because the trend over ten lots beats any single measurement.

Push what you cannot do outward. Require pre-shipment inspection at the supplier for offshore material, reported before the deposit balance is released. Send an annual sample to an accredited lab for properties you cannot measure, such as resin composition or flammability rating. And physically check anything from a supplier showing the warning signs in red flags when choosing a supplier.

As volume grows, connect incoming data to production data. When a molding process drifts, the first question should be which resin lot is in the hopper, and answering that in thirty seconds is what separates a controlled process from an argument. That is the practical payoff of running statistical process control on the line.

Set Up the Gate Before the First Shipment

Projects House writes IQC plans for hardware companies going into their first production run: part classification, inspection instructions, sampling levels, supplier scorecards, and the nonconformance procedure that keeps bad lots off the line. Send your bill of materials and supplier list through our contact form.