You approved FAI, released production, and still ended up with drifting dimensions, fit failures, or rejected batches. That sequence is common — and it’s not a contradiction.
It means the supplier could make a compliant sample, but never established a stable, repeatable production process capable of holding control at volume.
The sections below explain how to identify whether this failure is recoverable, what evidence proves loss of control, and when continuing production becomes a higher risk than switching suppliers.
Table of Contents
Why can parts pass FAI but still fail in production runs?
Parts pass FAI but fail in production because FAI proves a supplier can make one compliant sample — not that they can repeat it at scale. Many suppliers tune offsets, feeds, or inspection focus to pass the first article, then remove those controls once production begins.
FAI parts are often produced under ideal conditions: fresh tools, slower cycle times, experienced operators, and extra manual checks. When production ramps up, tool wear, cycle optimization, operator changes, and reduced inspection frequency expose weaknesses that were never controlled. Features with tight positional tolerances, coaxial relationships, thin walls, or post-processing dimensions are typically the first to drift.
A production-ready supplier treats FAI as the starting point of validation. That means locked tool paths, defined tool-wear limits, in-process inspection checkpoints, and proof that multiple consecutive parts meet spec — not just a single pass.
When those controls don’t exist, production failure is predictable.
Production Takeaway:
If parts failed after FAI, don’t approve another run unless the supplier can prove repeatability with production data — not assurances. No locked process, no in-process inspection history, no evidence of control means the process was never production-ready. Continuing at that point is risk, not progress.
Which production failures indicate the supplier never had a stable process?
Some failure patterns clearly show the supplier never had a stable process, even if FAI passed. The strongest signal is repeatable drift in the same features, especially failures that worsen over time instead of appearing randomly.
Common examples include bore diameters tightening as tools wear, positional tolerances shifting after fixture re-clamping, flatness changing after stress relief, or coated dimensions failing without compensation. These are not inspection misses — they are uncontrolled variables.
Another red flag is dependence on temporary fixes. If quality only recovers after offset resets, feed reductions, or part sorting, the process is not statistically capable. Early parts may pass, later parts fail, and quality depends on constant manual intervention rather than control.
Stable suppliers can show in-process inspection records, control limits, and defined reaction plans. Unstable suppliers can only promise that the next batch will be better.
Production Takeaway:
Patterned failures mean the process was never under control. If the supplier cannot show how variation is measured, limited, and corrected during production, stopping the run early usually prevents larger schedules and cost losses later.
Did the supplier treat FAI as a one-time setup instead of a production-ready process?
Yes — in most post-FAI failures, the supplier passed FAI under special conditions that were never enforced during production. FAI was treated as a one-time success, not a production baseline.
When we review these cases, the first thing we check is whether FAI conditions were identical to production conditions. In many failures, they aren’t. FAI is passed using slower feeds, fresh tools, senior operators, and manual offset tuning that disappears once volume machining begins.
A production-ready supplier can prove that the same tooling, programs, offsets, inspection points, and wear limits used during FAI are locked and enforced during production. When that documentation doesn’t exist, FAI approval has no predictive value.
Production Takeaway:
If the supplier cannot prove the FAI setup is identical to the production setup, FAI approval does not represent production capability. This is one of the first signals we use to judge whether a process is recoverable or fundamentally unstable.
FAI Passed — But Is This Process Actually Stable?
Passing FAI doesn’t guarantee production control. Upload your drawing and run data to see if this failure is recoverable.
Which dimensions or features begin drifting after FAI approval?
The features that drift after FAI are almost always wear-sensitive, fixture-dependent, or affected by secondary operations. These failures are patterned, not random.
In reviews, we consistently see drift in bore diameters, true position, coaxiality, flatness, thin-wall geometry, and dimensions affected by coating or heat treatment. These features pass FAI when tools are fresh and inspection is tight, then degrade predictably as production continues.
When the same features fail repeatedly, it tells us the process was never designed to control those variables. Random defects suggest execution issues; repeated drift points to missing process control.
Production Takeaway:
If the same dimensions keep failing after FAI, the process was never capable of holding them in production. This pattern is a primary indicator we use to assess whether continuing production will stabilize or continue to degrade.
Can the failed production parts be reworked—or are they scrap?
Some failed production parts can be reworked, but only if the failure was caused by correctable drift rather than inherent process instability. Many post-FAI failures are not recoverable.
In our assessments, parts failing due to offset creep or early tool wear may be recoverable once controls are established. Parts failing due to positional error, geometric distortion, or post-processing effects usually are not.
Rework becomes a red flag when it replaces root-cause correction. If quality only improves after offset resets or selective machining, the supplier is compensating for an unstable process rather than fixing it.
Production Takeaway:
If rework is required to meet spec, the process is already failing. Before approving rework, the supplier must show what process control changed — otherwise rework only delays scrap and increases loss.
When does a production failure invalidate the original FAI approval?
A production failure invalidates the original FAI approval once repeatability is broken and the supplier can no longer show the production process still matches the approved FAI conditions.
FAI approval is conditional. It only remains valid if the same tooling, programs, offsets, inspection points, and control limits used during FAI are still enforced during production. Once uncontrolled drift appears — or once the supplier changes how parts are made to “get them back in spec” — the approved FAI no longer represents reality.
In our reviews, FAI is effectively invalid the moment failures require offset resets, manual correction, selective inspection, or rework to pass. At that point, the process that produced the approved first article no longer exists, even if the paperwork still says “FAI accepted.”
FAI doesn’t fail on paper — it fails when the process behind it changes.
Production Takeaway:
If production no longer runs under the same conditions that passed FAI, the approval is no longer valid. Continuing without re-validation is not production — it’s exposure.
What evidence shows the supplier lost process control after FAI?
A supplier has lost process control after FAI when variation appears without defined limits, tracking, or documented reaction, even if only a small number of parts have failed.
The strongest signals are missing or inconsistent in-process inspection records, widening dimensional spread over time, undocumented offset changes, or quality “recovering” only after manual intervention. Another clear indicator is when the supplier cannot show when drift began — only that it exists now.
A controlled process produces trends, limits, and repeatable responses. An uncontrolled process produces explanations, resets, and assurances that things are “back to normal.”
When objective control evidence is missing, control should be assumed lost — not temporarily disrupted.
Production Takeaway:
If the supplier cannot show how variation is monitored and constrained during production, process control is already gone. Decisions should be based on data availability, not failure count.
Which production data proves a supplier can continue after failure?
To justify continuing production after failure, the supplier must provide objective evidence that the process is now stable under normal production conditions, not statements that corrections were made.
At minimum, this means in-process inspection records on critical features, defined control limits, documented corrective actions, and proof that multiple consecutive parts now meet spec without manual tuning. One corrected part or a single clean report is not evidence of recovery — it’s another FAI.
What matters is whether the supplier can show the process behaving predictably again, not whether they believe it will.
Production Takeaway:
If corrective actions are not supported by repeatable production data, there is no technical basis to continue the run. Stopping early is often the only defensible decision.
When should you stop production even if FAI was previously accepted?
You should stop production as soon as failures show the process is no longer repeatable and the supplier cannot prove control with objective data.
FAI approval does not justify continuing a run once production behavior changes. If parts only pass after offset resets, selective inspection, rework, or slowed cycle times, the process that earned FAI approval no longer exists. Continuing production under those conditions compounds risk instead of reducing it.
In real recovery cases, the costliest mistakes happen after the first failure—when teams approve “just one more batch” without demanding proof of stabilization. At that point, stopping is not escalation; it is containment.
Production Takeaway:
If production cannot continue under the same controlled conditions that passed FAI—and without corrective data to prove stability—the lowest-risk decision is to stop the run before losses multiply.
Still Running Parts After Production Failure?
Unstable production compounds scrap quickly. Share part and run data, decide whether to stop, recover, or switch suppliers.
Should you restart production with the current supplier or switch to a new one?
You should only restart with the same supplier if they can prove the root cause is fully corrected and the process is now stable under normal production conditions.
Restarting makes sense when failures were caused by a specific, documented issue—such as tool wear limits or fixture instability—and the supplier can show what changed, how it is controlled, and why it will not recur. Without that proof, restarting is based on trust rather than capability.
Switching suppliers becomes the rational choice when failures reflect missing controls, undocumented adjustments, or an inability to demonstrate repeatability. In those cases, restarting with the same supplier simply repeats the same risk under schedule pressure.
Production Takeaway:
If recovery depends on promises instead of measurable process changes, restarting is not recovery—it’s repetition. Switching suppliers is often the only way to regain control.
What info determines whether you should switch suppliers or retry production?
The decision to retry or switch depends on whether the supplier can provide objective evidence of process stability, not on how confident they sound.
Critical information includes proof that the production setup matches the approved FAI, in-process inspection data on critical features, documented corrective actions, and evidence that multiple consecutive parts now meet spec without manual intervention. What matters is not intent, but demonstrated behavior.
If that information is incomplete, delayed, or inconsistent, the technical basis for retrying does not exist. At that point, switching suppliers is not a reaction—it is a defensible decision based on risk control.
Production Takeaway:
If the supplier cannot prove stability with data, retrying production is a gamble. Switching suppliers becomes the only option that reduces schedule, cost, and quality risk.
Conclusion
FAI passing does not guarantee production success — process control does. When failures appear after approval, decisions must be based on data, not reassurance. Upload your drawing and inspection results for an independent production-stability review — we’ll assess recoverability and next steps within 24 hours.
Frequently Asked Questions
As soon as production fails after FAI and the supplier cannot provide objective evidence of restored control. An independent review helps determine whether recovery is realistic or whether switching suppliers is the lower-risk option.
Rework is acceptable only if it follows a proven corrective action that restores process stability. If rework is used repeatedly to meet spec, it masks the root cause and delays an inevitable stop decision—often at higher cost.
One repeatable failure on a critical feature is enough. When the same dimension drifts or fails again after adjustment, it signals a process control issue—not an isolated defect. Waiting for more failures usually increases cost without improving clarity.
Only if the original process remains unchanged and corrective actions are fully documented and controlled. If tooling, offsets, inspection methods, or cycle conditions change, the original FAI no longer represents production and must be revalidated.
No. FAI only confirms that one part met spec under specific conditions. It does not prove repeatability, tool-life stability, or process control at volume. Production capability must be demonstrated with in-process data and consistent results across multiple parts.
Recoverable drift is corrected by defined controls and verified with production data. A broken process relies on manual adjustments, rework, or selective inspection to pass. If stability depends on intervention, the process is not production-ready.
