Every quality engineer has inherited a stack of supplier submissions with a deadline attached and a manager asking whether they can be turned around today. The temptation is real: the paperwork looks complete, the index tabs are in order, and the supplier has signed everything. But approval is a technical act of engineering judgement, not a clerical exercise in confirming that pages exist.

A submission review done properly is a design review of the supplier's process, compressed into a few hours, and it deserves the same rigour. Across two decades of automotive and aerospace quality work, I have seen the consequences of skipping it: approvals that dissolved at the first audit question, and field failures traced back to evidence that described a process that never ran.

This article sets out a review method built on four disciplines: verifying that the evidence describes the real manufacturing route, sampling the technical substance instead of recounting it, matching reviewer competence to the commodity, and documenting the rationale so the decision survives scrutiny years later.

What Approval Actually Means

When you sign an approval, you are making a specific statement: this supplier can produce this part, on this process, repeatably, and the evidence is traceable to the real manufacturing route. That last clause matters more than most reviewers acknowledge. I have reviewed submissions where the dimensional results came from a prototype cell run by the development team, on a machine that would never see serial production. The parts were beautiful. They were also irrelevant.

So the first question in any review is whether the evidence describes the process that will actually run. Check the machine identifiers in the process flow against the capability study. Check tooling numbers against the tooling register. Check whether the material certificates reference the actual heat lot used for the samples, or a representative lot supplied afterwards. Traceability breaks in these small gaps, and every gap is a place where a rubber stamp hides.

Approval also means you accept the supplier's measurement method, their sampling logic, and their definition of a conforming part. If you cannot articulate, in your own words, how the supplier decides a part is good, you have not reviewed the submission. You have filed it.

The evidence chain you must trace

  1. 01Chemistry certificateMill certificate for the actual heat lot, not a representative one
  2. 02Incoming inspection recordSupplier verified the lot on receipt, with dates that fit the timeline
  3. 03Process recordsFurnace charts and heat treatment covering the claimed soak times
  4. 04Sample identificationRetained samples tied to the lot and to the results reported
  5. 05Dimensional resultMeasured on the serial production route and tooling
Trace one lot forward through the package; any break in the chain is grounds for a full review of that thread, because traceability errors cluster.

Checking the Technical Substance

Start with drawing interpretation. Suppliers sometimes measure what is easy rather than what is functional. A housing with a critical bore position relative to a datum scheme will often arrive with radial runout results instead, because runout is simpler to fixture. That is not malice; it is a measurement convenience that quietly changes the acceptance criterion. Compare the characteristic list against the drawing's datums and note every substitution.

Then interrogate the material and process evidence. Look at hardness results across the sample set, not just the averages, because spread tells you whether the furnace or the heat treatment subcontractor is stable. Examine weld documentation for penetration evidence where the joint is structural, and request macro sections if none were supplied. For castings and forgings, porosity and grain-flow documentation should correspond to the X-ray or witness locations defined on the drawing, not to a generic inspection plan.

Surface treatment thickness, torque-tension relationships on fasteners, cleanliness results for hydraulic components: each product family has its own technical anchors. A reviewer who knows the family can smell a weak submission within minutes. The reviewer who does not will default to checking that documents exist, and that is precisely how rubber-stamping begins.

Approval decisions are made at the supplier's process, not in the paperwork that describes it afterwards.
Approval decisions are made at the supplier's process, not in the paperwork that describes it afterwards.

Sampling the Evidence Rather Than Recounting It

You cannot re-measure every dimension on every sample, and you should not try. Sample intelligently instead. Pick three to five characteristics: one that is functionally critical, one that historically causes trouble for this process family, one where the supplier's own results sit close to the limit, and one measured with what looks like the least capable method. Verify those independently, either by sending retained samples to your own metrology or by witnessing the supplier re-measure them.

Discrepancies between your numbers and theirs tell you more about the supplier's quality system than any audit checklist. The same logic applies to documentation: rather than reading every certificate, pick one material lot and trace it forward from chemistry certificate through heat treatment record to the dimensional result. A date that does not fit, a signature from someone who left the company before the samples were made, a furnace chart that does not cover the claimed soak time — each is grounds for a full review of that thread.

Resist the urge to recount what the supplier already counted. If their results table lists forty dimensions with values, do not re-verify the arithmetic. Ask instead whether the reported measurement uncertainty supports the conclusions drawn, and whether the number of samples justifies any stability claim. Sampling your own verification effort is what makes review sustainable at industrial volume.

If you cannot articulate how the supplier decides a part is good, you have not reviewed the submission. You have filed it.

Reviewer Competence and the Courage to Say No

A submission review requires a reviewer who understands the manufacturing process behind the paperwork. Reviewing a deep-drawn stamping submission without knowing what wall thinning, springback and burr formation look like in measurement data means you are reviewing form, not content. Reviewers for a given commodity should have either worked the process or completed structured training on its failure modes. Where neither is feasible, review in pairs: one metrology-competent reviewer, one process-experienced engineer.

Competence includes knowing your own limits. A good reviewer writes I cannot assess the validity of this coating adhesion data and pulls in a surface-treatment specialist, rather than approving with a silent mental asterisk. In my experience the worst approvals are not the lazy ones; they are the ones made by a competent generalist too proud to admit the specialist gap.

There must also be an organisational route to rejection that carries no penalty. If rejecting a submission means a difficult phone call from purchasing every time, rejection rates will drop regardless of submission quality. I have made it explicit policy in my teams: a rejection with documented technical reasons is a completed review, treated identically to an approval in performance terms. Reviewers who feel safe saying no say it earlier, when problems are cheap.

Common Failure Modes in the Review Itself

Watch for the perfect submission: no anomalies of any kind, every dimension at mid-tolerance, every test result comfortably identical. Real manufacturing has texture. Suspiciously uniform data can indicate selective reporting, reworked samples, or measurements taken on a golden part. Ask for the raw output — the CMM probe paths, the printouts, the raw hardness indent readings. Suppliers with genuine data produce it without drama.

A second failure mode is approving against a superseded specification. Drawings get revised; submissions get prepared against the old revision and submitted late. The revision level on every document in the package must match the current drawing and any open engineering change requests. I have caught this more often than any other single error, always the same boring way: a deliberate, line-by-line revision check at the start of the review, before anything technical happens.

The third is the approved-previous-time fallacy: we approved their last three submissions, so this one is fine. Process changes, tooling refurbishment, personnel turnover and material source changes can all degrade a supplier between submissions. Each package stands on its own evidence. Previous approvals establish trust in the relationship, never in the part.

Rubber-stamping versus real review

Filing the submission

  • Confirms every required document exists
  • Trusts reported values as printed
  • Approves against whichever revision arrived
  • Skips unfamiliar technical sections
  • Records an outcome and a signature

Reviewing the submission

  • Traces evidence to the serial production route
  • Independently re-measures sampled characteristics
  • Checks all revisions against the current drawing
  • Escalates specialist gaps explicitly
  • Documents the rationale and open points
The two modes look similar in the tracker — both end in a signature — but only one survives an audit or a field failure.

Structuring the Review for Defensibility

Whatever you approve, you must be able to defend it two years later, when a field failure occurs and someone asks who reviewed the submission and on what basis. Document not just the outcome but the rationale: which characteristics you verified independently, which documents you traced, what questions you asked the supplier, and what their answers were. A one-page review record with that content is worth more than a signed cover sheet with forty tick boxes.

Define decision rules before the review, not during it. Which deviations are acceptable with a documented concession, and which force a resubmission? What constitutes a major finding versus a minor one? Written criteria remove the negotiation element and make approvals consistent across reviewers, which protects both the supplier relationship and your own position when a decision is later questioned.

Finally, close the loop. When a submission is approved with open points, those points carry owners and dates, and the approval record references them. When a submission is rejected, the rejection reason goes back to the supplier in technical language specific enough to act on. Reviews that end in silence breed repeat submissions of the same defective package, and the second review is always harder than the first properly executed one would have been.