Inspection timing is the highest-leverage quality decision a buyer makes, because the value of a finding decays as production advances. An initial production inspection booked when the first representative output exists, and the bulk of the run still lies ahead, converts process faults into corrections instead of disputes. This guide shows how to read the signals, trust the sampling evidence, and schedule the visit where it changes outcomes.
Key takeaways
- The 10 to 30 percent window balances representativeness against correctability, and regulated or complex orders belong at its earlier end.
- New suppliers, changed inputs, complex assemblies, and tight schedules are the four signals that make early booking mandatory rather than optional.
- Sampling reasoning shows systematic process faults reveal themselves in modest samples, justifying immediate correction[1].
- Acceptance testing logic frames the visit as a gate: verify the process against criteria before letting volume accumulate[2].
- Waiting converts correctable process findings into finished-goods disputes, trading a cheap early fix for an expensive late negotiation.
Every buyer faces the same scheduling question: inspect early while correction is cheap, or wait until delivery when the facts are complete but the leverage is gone. The answer depends on reading the order risk signals correctly: the supplier history, the stability of materials, the complexity of the product, and the strictness of the schedule.
This article develops that reading skill. It explains the process signals that argue for early booking, shows how sampling reasoning supports acting on modest early samples, applies acceptance testing logic to the timing decision, and addresses the cost objection that keeps many buyers waiting too long.
The setting is familiar: an order with a supplier the buyer cannot watch daily, goods that must match an approval, and a shipment date that will not move. Within that setting, timing is not administrative detail but the central quality decision, because the same finding is worth ten times more at 20 percent completion than at 100 percent.
Reading the four signals that demand early booking
New suppliers top the list because no history exists on how they interpret specifications. Drawings, samples, and written instructions always leave gaps that experience fills, and a new factory fills those gaps by guessing. Some guesses are right; the early visit exists to catch the wrong ones before they scale. First orders with unfamiliar factories should therefore default to the earlier end of the inspection window, around 10 to 20 percent completion. UTS Inspection
Changed inputs form the second signal. A new material batch, a substituted component, a revised artwork file, or an updated specification reintroduces uncertainty into an otherwise familiar process. The factory habits were tuned to the old input, and the new one behaves differently in ways nobody predicted. Any order carrying a change since the last successful run deserves an early visit focused on exactly what changed.
Complexity and schedule pressure complete the set. Complex assemblies multiply the cost of each misunderstanding because one wrong sub-component infects every finished unit, while tight schedules remove the rework buffer that late discovery needs. When either is present, the early visit functions as schedule insurance. Service teams at UTS Inspection are frequently scoped around these signals, with criteria weighted toward whatever the order risk profile makes most expensive to discover late.
Why the 10 to 30 percent window is the sweet spot
Below roughly 10 percent completion, the line may still be settling, with trial settings, first material threading, and operators finding rhythm, and the sample reflects startup noise as much as process truth. Findings from that stage can mislead, triggering corrections to problems that would have resolved themselves. Above roughly 30 percent, the arithmetic turns against the buyer: each additional day of uncorrected production adds finished units that must be reworked individually rather than fixed by one process change.
Within the window, position follows risk. Simple repeat orders with proven suppliers can sit toward 25 to 30 percent, where the sample is large and stable. New suppliers, new designs, and regulated goods belong toward 10 to 20 percent, where even a modest correction lead time matters enormously. The buyer need not hit an exact number; the discipline is booking against the milestone with a fallback day.
One common mistake is treating the window as a deadline rather than a target: rushing a visit at 8 percent to tick a box, or postponing past 40 percent because the line started late. Both errors waste the visit economics. A slightly late visit with a representative sample still beats an on-calendar visit to an unready line, so buyers should move the date with the milestone rather than defending the original plan.
Trusting small samples for systematic faults
The hesitation to act on a few dozen units from an order of tens of thousands is natural but, for systematic faults, misplaced. Acceptance sampling theory distinguishes sharply between random variation, which needs large samples to quantify, and systematic process causes, which announce themselves almost immediately[1]. A wrong thread, an incorrect logo plate, or a mis-set die appears in nearly every piece the process produces. When most of the early sample shares one fault signature, the buyer is observing the future of the whole order, not a run of bad luck.
The practical skill is distinguishing the two cases in the report. Systematic findings repeat with one signature across units; random findings scatter across fault types and locations. The early report should label each finding accordingly, because the responses differ completely: systematic faults demand root-cause correction and a line stop, while random faults merit monitoring and in-process control.
There is an honest limit to state as well. A clean early sample does not certify the unseen remainder of the order: materials can change, tooling can wear, supervision can relax. The early visit diagnoses the process at its start; continued assurance needs either stable controls or follow-up checks. Presenting early results as a final verdict overstates them. Presenting them as the baseline for monitoring states their value exactly.
The acceptance gate applied to production
Capital equipment buyers long ago formalized the timing intuition into factory acceptance testing: before committing to installation and payment, verify defined checkpoints against agreed criteria, record the results, and proceed only when the evidence supports it[2]. The logic transfers directly to volume production. The early visit is the acceptance test of the running process, with first output examined against specification and sample, deviations documented, and volume continuing on the strength of that evidence rather than on hope.
Framing the visit as a gate changes the buyer internal conversation. Instead of asking whether inspection is worth its fee, the buyer asks whether unmonitored volume is worth its risk: tens of thousands of units produced by a process whose capability nobody verified. For high-risk profiles the answer is plainly no, which makes the visit self-justifying. For low-risk repeats the answer may be yes, and skipping the early visit is then a reasoned decision rather than a habit of postponement.
The gate needs teeth to work. Findings must trigger defined consequences, with root-cause responses within days, re-verification before the halfway mark, and escalation if corrections fail, or the gate is decorative. Buyers should state these consequences in the briefing so the factory meets the visit as a milestone with known outcomes. A gate whose failure nobody defined in advance produces the worst of both worlds.
Answering the cost objection honestly
| Cost element | Fault caught at 20 percent | Same fault caught at 100 percent |
|---|---|---|
| Rework scope | Process fix plus small-batch correction | Open, rework, and repack finished cartons |
| Material consequence | Swap incoming stock before consumption | Material already built into every unit |
| Schedule consequence | Absorbed within production buffer | Missed vessel, expedited freight, penalties |
| Negotiation position | Factual correction during production | Dispute over finished goods under deadline |
| Evidence quality | Dated process photos from the line | After-the-fact claims without process record |
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
Frequently asked questions
What is the ideal completion percentage for an early inspection?
Book when roughly 10 to 30 percent of the order is complete: enough output exists to examine representatively, while most production still lies ahead for corrections to take effect. Earlier than that, the sample may not represent the running process; later than that, corrections grow expensive as finished volume accumulates. Regulated or complex orders should aim at the earlier end of the window.
Which signals mean the early visit should be scheduled sooner?
Three signals argue for earlier scheduling: a new supplier with no track record, any change in materials or specifications since approval, and complexity that multiplies the cost of a single misunderstanding. A fourth signal is schedule pressure, since tight launches leave no rework buffer. When two or more signals coincide, treat the early visit as mandatory rather than optional.
Can a small early sample really justify stopping the line?
Yes for systematic process faults, which is exactly what early visits target. When a defect comes from a process cause, such as a wrong setting, wrong material, or misunderstood instruction, it appears in nearly every piece, so a modest sample proves the problem is real and line-wide. Random variation is a different matter and needs larger samples to quantify, but random variation is not what early visits are for.
How does acceptance testing logic apply to inspection timing?
Acceptance testing practice says verify defined checkpoints against agreed criteria before commitment deepens, recording results as evidence. Applied to production, the early visit tests first output against the specification and golden sample, files the readings, and lets volume continue only when the evidence supports it. That gate function is the timing logic in its purest form.