During-production inspection is worth the cost when finding a repeatable defect now could change the remaining production, and your supplier still has time to act. The usual UTS DPI window is 30%–50% complete, leaving 50%–70% unmade—not automatically protected. For a stable, low-risk repeat order, final inspection alone may be proportionate. [1][2]
We are UTS, a third-party inspection company providing product inspection, factory audits, supplier evaluation, and laboratory-test coordination. We assess the value of a mid-production visit through three questions: what could go wrong, how much production remains, and what can still change before shipment. [3]
What Does a During-Production Inspection Add?
A during-production inspection (DPI), also called DUPRO, checks available finished output while the order is still running. Unlike a final random inspection, it offers an earlier view of workmanship, dimensions, function, packaging, and labels against your approved requirements. Our DPI service normally takes place at 30%–50% completion; final random inspection takes place when production is complete and at least 80% is packed. [1][2]
The difference is timing, not a promise of defect-free goods. A DPI finding may help you and your supplier address a recurring issue before more units are made. It does not verify goods that do not yet exist, replace final verification, or establish legal compliance by itself.
How Our DPI Process Makes the Visit Useful
Our published DPI process starts with agreeing the checklist: specifications, the approved sample, defect definitions, and acceptance levels. We then select samples, conduct the agreed checks, classify observed defects, and prepare the inspection report. This sequence matters: without a clear reference, a mid-run observation may not answer whether production meets your requirements. [1]
We compare sampled goods with the agreed specifications and approved reference sample. Our service page states report delivery within 24 hours after inspection and deployment within 48 hours; confirm site access and availability with our team before relying on either timing for a particular order. [1][4]
Our reports present actual findings only: the inspection scope, measurements, observations, test results, defects, and supporting photographs. They do not include corrective-action, improvement, or shipment-release recommendations. Decisions about changes and shipment remain with you and your supplier.
When Is a DPI Most Valuable?
A Repeatable Defect Could Affect a Long Run
Consider a hypothetical 10,000-unit order inspected at 40% completion: 4,000 units are made and 6,000 remain unmade. If the sample reveals a recurring assembly problem, there may still be an opportunity to change the process for those remaining units. The 6,000 figure describes production exposure—not units guaranteed to be saved. Existing output may also need separate assessment.
The Supplier, Materials, or Process Has Changed
A previously approved sample does not establish that a changed material batch or production setup will perform identically. A DPI can check sampled output after a change or an earlier finding. If the concern is supplier capability rather than this run's output, start with factory evaluation and supplier verification; the two services answer different questions. [5]
The Schedule Still Allows a Response
A tight deadline makes earlier information useful only if there is still time to use it. Confirm when the findings will arrive, whether the supplier can adjust the process, and whether any necessary verification fits before shipment. If production will finish before anyone can respond, the value of a mid-run visit is reduced.
The Order Contains Several SKUs
Sample allocation matters. In a hypothetical equal split, 200 samples across 10 SKUs gives 20 per SKU; that arithmetic is not an approved sampling plan. Agree coverage by SKU, product risk, lot structure, and available inspection time with our team. There is no universal fixed minimum per SKU for every mixed order. [6]
When Might You Skip or Reschedule the DPI?
A small, simple repeat order with consistent documented results and no material or process changes may justify final inspection rather than an additional visit. That is a risk-based choice, not proof that the new batch is unchanged.
If the concern is incoming materials or initial setup, an initial production inspection at the 5%–10% stage may be more appropriate. If the order is nearly finished, compare what a late visit could still establish with what final inspection will cover. There is no universal 70%–80% cutoff at which an inspection becomes worthless. [7]
How Much Assurance Does the Sample Provide?
Acceptance sampling uses a sample to inform a lot decision; it does not establish that every unit is defect-free. ISO 2859-1 describes AQL-indexed sampling schemes for lot-by-lot inspection. Agree the applicable standard, inspection level, lot definition, and plan before treating any sample count as a requirement. [8][9]
Our published sampling examples include a 4,000-unit lot at General Inspection Level II, code L, with a 200-unit sample and acceptance/rejection counts of 10/11 at AQL 2.5 under the stated single-sampling plan. Another example uses 80 samples for a 1,000-unit lot. These are plan-dependent examples, not automatic rules for every DPI or every SKU. [10][11]
For the 4,000-unit example, 200 samples are 5% of the lot. AQL 2.5 does not mean inspecting 2.5% of the order, nor does passing the sample prove the whole order contains no more than 2.5% defective units. [8][10]
How to Decide Whether the Extra Visit Is Worthwhile
Inspection scope depends on the SKU mix, measurement and function-test requirements, test setup, location, and access to the goods. Agree those requirements before comparing a mid-production visit with final inspection alone.
| Decision question | A stronger case for DPI | A weaker case for DPI |
|---|---|---|
| Could the same issue recur? | Changed inputs, unresolved findings, or a long remaining run | Stable, simple repeat production with documented results |
| Can anyone act in time? | The supplier can respond and verify changes before shipment | Production will finish before findings can be used |
| Will the visit answer a defined question? | Agreed checks and meaningful coverage of the relevant SKUs | No clear checklist or access to representative goods |
We would not treat order size alone as the deciding factor. A small safety-critical order may warrant closer control than a much larger, straightforward repeat run. The practical comparison is the value of earlier evidence against the additional inspection work—not a fixed percentage of production remaining.
What to Prepare Before the Visit
Send our team quantities and progress by SKU, current specifications, the approved sample, defect criteria, and the required tests. Identify the particular question the DPI must answer, such as whether a previously observed dimensional issue is present in current sampled output. Our published process allows buyer-specific requirements to be incorporated into the checklist. [1]
For barcode or QR checks, agree the sample scope, scanning device, test conditions, and expected encoded data beforehand. 100% of the tested codes must scan successfully and match the approved data. This is a requirement for the tested objects, not a claim that the whole lot was scanned or that every device will read them later. For linear-barcode print-quality assessment, ISO/IEC 15416 provides the relevant method; it is not a QR-code grading standard. [12]
The Bottom Line
A DPI is most useful when it answers a specific quality question while production can still respond. Use the 30%–50% window to plan the visit, not as a stand-alone rule for whether it is worthwhile. Keep final verification in view: an earlier sample cannot confirm the quality of later output. [1][2]
Would a mid-production visit answer your quality question?
Send our team your SKU mix, production progress, specifications, and shipment date. We can review the timing and agree what the inspection will cover.
Contact UTS about your DPISources
- UTS During Product Inspection service page — https://www.utsinspection.com/during-product-inspection.html
- UTS Final Random Inspection page — https://www.utsinspection.com/final-random-inspection.html
- UTS company homepage — https://www.utsinspection.com/
- UTS Sample Evaluation page — https://www.utsinspection.com/sample-evaluation.html
- UTS Factory Evaluation and Supplier Verification page — https://www.utsinspection.com/factory-evaluation.html
- UTS Product Quality Inspection for Multi-SKU Orders — https://www.utsinspection.com/news_view_100.html
- UTS Initial Production Inspection page — https://www.utsinspection.com/initial-production-inspection.html
- NIST/SEMATECH e-Handbook of Statistical Methods, acceptance sampling — https://www.itl.nist.gov/div898/handbook/pmc/section2/pmc21.htm
- ISO 2859-1:2026, Sampling procedures for inspection by attributes, AQL-indexed schemes for lot-by-lot inspection — https://www.iso.org/standard/2859-1
- UTS Pre-Shipment Inspection Cost Guide — https://www.utsinspection.com/news_view_121.html
- UTS How AQL Sampling Works in Third-Party Inspection — https://www.utsinspection.com/news_view_53.html
- ISO/IEC 15416:2025, Bar code print quality test specification for linear symbols — https://www.iso.org/standard/83390.html
