Mis-shipment Prevention | 5 Gates Where Shipping Errors Are Created

You installed a shipping verification system, and yet mis-shipments still have not gone to zero. It is a story we hear constantly at Japanese-owned factories in Thailand. Every item is scanned and matched against the order, and still a few wrong deliveries come back from customers each month. The shop floor responds with “we will inspect more strictly,” more checkers are added, dispatch slows down, and the numbers barely move. Mis-shipment prevention stalls at this exact point because most errors are not created at the moment of shipping. They are created upstream. This article breaks the process into five gates and draws the line between the errors barcode verification actually stops and the errors that pass through verification perfectly intact.

Why mis-shipment prevention fails when it starts with “stricter inspection”

The post-incident review meeting follows the same script in most factories. The root cause is recorded as “insufficient checking,” the countermeasure is written as “enforce double-checking,” and a corrective action report goes to the customer. When a different mis-shipment happens the following month, “triple-checking” gets added. Six months later there are three people standing at the dispatch area, and the mis-shipment count is almost unchanged.

The reason for this dead end lies in the nature of verification itself. What barcode verification judges is exactly one thing: whether the information written on the shipping instruction matches the label attached to the physical item. If the instruction itself is wrong, verification returns a clean match against the wrong instruction. If the item tag does not describe what is actually in the box, verification returns a clean match against the wrong label. In both cases the system works exactly as designed and lets the mis-shipment through.

In other words, inspection was never built to detect errors created upstream. “Inspect more strictly” is effective against the class of errors it can stop, but against the class it cannot stop, it consumes labor and time and produces nothing. Worse, adding checkers dilutes individual accountability and creates a state where everyone assumes someone else is watching. There is a range where adding people actually lowers accuracy, as anyone who has run a shop floor knows from experience.

So the first thing to do in mis-shipment prevention is not to tighten inspection. It is to identify which gate is producing your own mis-shipments. Once you know the gate, the countermeasure follows. Strengthening inspection without identifying the gate means permanently staffing people against errors they cannot possibly catch.

Take your past mis-shipments one by one and reclassify them by where the error entered the process. In most factories, errors that shipping inspection can never stop by design will account for a far larger share than expected.

The five gates where mis-shipments are created

Shipping is not a single operation. You receive an order, create a shipping instruction, allocate stock, locate and pick the physical item, pack it, apply a label, and load it. Across that chain there are five gates where an error can enter.

Gate 1 Order entry and shipping instructions — errors inspection can never stop

This is the most upstream gate. Errors created here pass cleanly through every downstream process.

The typical error is a keying mistake in part number, quantity, or delivery destination. If customer orders flow in automatically via EDI, keying errors do not happen. But when orders arrive as PDF attachments, Excel order forms, or in some operations still by fax, someone is retyping them into the core system. At that moment of transcription, a quantity with a dropped digit or a part number off by one character enters the system.

Another common case is shipping against superseded information. A design change switched the drawing number, but the old order for the old part number is still open, and the shipping instruction carries the old part number. The warehouse issues the old part exactly as instructed, and the barcodes match. Nobody inside the company notices until the customer reports that an obsolete part arrived.

Unit confusion is also an upstream error. When the customer orders in cases and internal stock is managed in pieces, an incorrect pack-quantity master produces a shipment inflated by the digit span of the pack quantity, and it is fully compliant with the instruction.

The countermeasure at this gate is not inspection. It is changing the structure of data entry. Ingest order data in machine-readable form. Hold units and pack quantities in the master so nobody types them by hand. Make revisions switch automatically by effective date. And place an anomaly detector at the shipping-instruction stage. Even a simple alert when the quantity jumps sharply versus the previous order, or when a delivery destination appears for the first time, changes how many errors get through.

Gate 2 Matching stock data to physical items — when the item tag lies, verification passes the lie

The second gate is whether the stock recorded in the system matches what is physically on the shelf. When this breaks down, the verification system simply approves the wrong label.

The classic shop-floor event is an item tag that does not describe its contents. Leftover material returned from a process was transferred into a different box while the original tag stayed on it. Similar parts were stored on the same shelf and their tags got swapped. A handwritten item tag had its part number misread. Scan that box with a handheld terminal and the part number printed on the label is read, it matches the instruction, and it ships. The error is inside the box, not on the label, so verification cannot see it.

Wrong locations have the same structure. If the shelf address does not match reality, something else is sitting where the picking list points. Where lot control applies, lot mix-ups are added on top. Issuing the wrong lot for a part under traceability requirements makes the recall scope impossible to define, and the damage no longer stops at a single mis-shipment.

The countermeasure belongs on the side that creates the match between item and label. Print and apply the label from the system at the moment of receiving. Define an operating rule that reissues a label every time material is transferred. Put barcodes on shelf locations as well so that a dual check of shelf and item can be enforced. Use cycle counting to find discrepancies early. Since this is also a question of raw stock data accuracy, design it alongside the thinking in our comparison of factory inventory management systems and optimal inventory level management so the investments do not overlap.

Gate 3 Picking — where barcode verification earns its keep

The third gate is whether the instructed item is correctly taken off the shelf. Of the five gates, this is arguably the only place where barcode verification delivers its full designed value.

The typical errors are look-alike part numbers: numbers differing only in the last character, color variants, right-hand and left-hand versions, old and new part numbers. Human eyes skim past differences in strings they see every day. Quantity errors are frequent too, and when nested units are involved, such as ten bags of ten pieces each, even veterans make mistakes. FIFO violations are, for date-controlled materials, effectively mis-shipments as well.

All of these stop on the spot when a handheld terminal matches the instruction against the barcode on the physical item. Requiring a location scan before picking also stops the error of taking from the adjacent shelf. Counting quantity by number of scans eliminates keyed digit errors. The operational design details are covered in picking with handheld terminals and barcodes.

What matters is not concluding that “we have a verification system, so we are covered” on the strength of the errors Gate 3 stops. Once this gate is hardened, the remaining mis-shipments shift in composition toward Gates 1, 2, and 4. Most factories that feel the reduction was smaller than expected are simply not watching that shift in composition.

Gate 4 Packing, enclosures, and piece counts — omissions slip straight past verification

The fourth gate is packing, and it contains a region where verification is structurally weak.

Barcode verification only sees what was scanned. What was never scanned, meaning what was left out, leaves no record at all. Accessories, spare parts, enclosed documents such as inspection certificates and mill sheets, one carton out of a multi-carton shipment. Any of these can be missing and the system still completes normally, because everything that was scanned matched.

The reverse error exists too. An item from a different order is sitting on the workbench and ends up in the same box. The contaminating item is not on the instruction, so it is never scanned, and again there is no record. Packing-unit errors, such as shipping in cardboard what should go in a returnable container, or stacking the wrong quantity per pallet, also escape line-item verification.

The countermeasure is to raise the unit of verification from the line item to the package. Clear order line items as the package is closed, and block the outer label from being issued until every line item is cleared. Omissions then stop at the label-issuing step. Weight verification works well in combination: reject any package where the difference between theoretical and measured weight exceeds tolerance. It will not work for small parts where the per-piece weight difference is negligible, but for items above a certain weight it reliably detects short counts.

Gate 5 Label application and destination — swapped in the final second

The last gate is whether the correct destination label goes on the correctly packed box. Everything upstream can be perfect, and a single misapplied waybill still counts as a mis-shipment from the customer’s point of view.

The conditions that produce it are almost always the same. Shipments for several destinations are staged side by side in the dispatch area. Labels are printed in a batch and carried around as a stack. Multiple printers are running and labels get picked up from the wrong one. Because the action takes only a few seconds, even the person who did it retains no memory of it, and root cause analysis goes nowhere.

The countermeasure is to abandon the “print in a batch and apply later” pattern. Scan the box you are about to label, then issue only that box’s label. With that sequence, misapplication becomes structurally impossible. Add redundancy on top, such as physically separating floor lanes by destination and scanning at pallet level during loading to match against the truck run, and the final-second error effectively disappears.

Lining up the five gates by type of error, whether shipping inspection stops it, and the effective countermeasure makes it clear where investment belongs.

Classifying your own mis-shipment history against this table can change your investment priorities. Starting from Gate 3 is the standard playbook, but if most of your history sits in Gate 1, fixing the structure of order entry first will do more.

Measuring where you stand with mis-shipment rate in PPM

Before choosing countermeasures, pin down your current level as a number. The common metric for logistics quality is the mis-shipment rate, normally expressed in PPM, or parts per million.

The formula is straightforward. Mis-shipment rate in PPM = number of mis-shipments divided by total transactions, multiplied by one million. One error in 100 transactions is 10000 PPM; one error in 10,000 transactions is 100 PPM. PPM is used instead of percentages because once you reach a good level, the percentage figure fills up with decimal places and becomes hard to compare.

The benchmark levels can be organized as follows.

A realistic sequence is to set the first target at breaking below 100 PPM, and from there aim for the 10 PPM range. A site sitting around 1000 PPM that jumps straight to fully automated verification of every unit will find the number moves far less than expected, because the upstream masters and item tags are not in order.

There are three cautions when working with these numbers.

First, the definition of the denominator. Whether you count total transactions as shipping documents, as line items, or as pieces changes the resulting figure by an order of magnitude for the identical reality. Fix the definition internally, and when comparing against published figures from other companies, confirm what they counted. If the definition is never written down and the responsible person changes, next year’s figure cannot be compared with this year’s.

Second, the definition of the numerator. If only customer complaints go into the numerator, everything caught internally and stopped before shipping disappears from view, and the effect of your improvements becomes invisible. Record cases caught internally, meaning prevention, separately from cases that reached the customer, meaning escapes, and you can read which gate is actually working.

Third, small population size. When monthly shipping line items number in the low thousands, a single mis-shipment moves the figure by hundreds of PPM. Rather than reacting to month-to-month swings, judge the trend on a quarterly moving average.

Three verification methods — visual double-check, handheld verification, full automated verification

The means of stopping errors fall broadly into three methods. Each differs in which gates it covers, what it costs, and how much burden it places on the floor.

The visual double-check has a person compare the instruction sheet against the physical item. It requires no additional investment and can start tomorrow. But it only stops differences a person can see. It is weak against look-alike part numbers and off-by-one-digit quantities, precisely the errors human eyes skip, and its accuracy depends on operator experience and physical condition. As shipping volume rises, checking hours rise linearly with it.

Handheld verification matches the instruction against the item’s barcode using a mobile terminal. It reliably stops Gate 3, and by extending scope to locations and outer labels it also covers part of Gate 2 and all of Gate 5. Implementation centers on terminals, label issuance, and system integration, and it involves changing work procedures. For most Japanese-owned factories, this is the option with the best balance of cost and effect.

Full automated verification checks every unit without human operation, using fixed scanners, weight verification machines, image inspection, or RFID gates. It does not slow throughput and does not depend on operator skill. On the other hand, capital investment is large, and it assumes consistency in the shape, material, and packaging of the target items. Applicability must be assessed carefully, since RFID reading is unstable near metal parts or liquids, for example.

In practice these three are not mutually exclusive. A common combination is handheld verification as the base with weight verification added only for items where piece count is critical. The important point is that no method stops Gate 1 errors. Treat the choice of verification method and the redesign of upstream data entry as two separate tracks running in parallel.

Who issues the item tags and labels changes the outcome

What ultimately determines verification accuracy is not scanner performance but whether the information printed on the label is correct. This is easy to underrate, yet whether Gate 2 errors occur at all is decided almost entirely by this single point.

The state to avoid is item tags and labels being created in several places, each by its own method. Receiving prints from a staff member’s Excel ledger, in-process uses handwritten cards, dispatch creates labels in yet another label application. Under those conditions the same part number exists in different notations, and every transcription creates room for error. Handwritten item tags are not only misread; they carry no barcode at all, so they cannot be verified in the first place.

The governing principle is simple: item tags and labels should be issued by the system that holds the inventory. Part number, lot, quantity, and pack quantity held in the inventory or production management system are passed straight to the label issuing system and printed. If you never create a step where a person retypes a number, transcription errors cannot occur.

Reissue control also has to be defined. Labels get dirty and torn, so reissues will always happen. If anyone can reprint something roughly on a nearby printer, two supposedly identical labels with different contents end up on the floor. Reissue must go through the system, and the old label must be recovered and destroyed. Simply writing those two rules down visibly reduces unexplained verification errors.

The physical quality of the label cannot be ignored in Thai factories either. Hot and humid warehouses, processes where cutting oil sprays, temporary storage in outdoor yards. When print smears, peels, or fades under those conditions, unreadable labels multiply, and the floor invents a workaround where anything unreadable is passed by visual check. The thinking behind barcode symbology and print quality is organized in selecting barcodes, QR codes, and RFID for traceability. Give label material and printer selection the same weight in your evaluation as the system design itself.

The “inspection-free receiving” customers ask for — SSCC and ASN

In the automotive parts and food industries, customers sometimes require deliveries in a form they can receive without opening the cartons. SSCC sits at the center of that mechanism.

SSCC, the Serial Shipping Container Code, is an 18-digit identifier assigned to each individual shipping package, such as a pallet or a case. It consists of an extension digit, the GS1 company prefix, a serial reference, and a check digit. Its application identifier is (00), and it is shown on the SCM label as a GS1-128 barcode. The essential point is that it identifies “this specific box” rather than a part number, so ten boxes of the same part number carry ten different SSCC values.

The code delivers its value when combined with an advance shipping notice, or ASN. The shipper sends, in advance, an ASN describing what and how much is inside each SSCC-labeled box. The receiver scans the outer label on the arriving box, matches it against the ASN, and completes receiving inspection without opening anything. Because receiving labor drops sharply, the customer has a strong incentive to demand it.

From the shipper’s side, this is also a mechanism for proving your Gate 4 and Gate 5 control to an outside party. If the ASN contents differ from the box contents, the receiver accepts without opening, and the error flows straight into their process. Supporting SSCC and ASN therefore presupposes that line-item clearing at packing time is reliably working. The natural sequence is to establish package-level control internally first, then give outer labels a serial identity, and only then move on to transmitting ASNs.

If there is any chance a major customer will require ASNs in the future, we recommend designing your verification system now with a structure that can hold a unique number per package. Changing the granularity of package control later causes considerably more rework than designing for it from the start.

Cost of a mis-shipment prevention system and phased implementation

Cost varies widely with the number of processes in scope, the number of item codes, the number of sites, and the state of existing systems. This article does not assert figures. Instead it sets out the cost components and the sequence in which investment should be made. When you get to actual quotations, comparing initial cost and annual running cost separately in Thai baht keeps the internal discussion from drifting.

There is a reason for this sequence. The handheld verification in Phase 2 only produces results once Phase 1 has made the label contents correct. Hand out terminals while labels still disagree with physical items, and verification passes while mis-shipments remain, leaving the floor with the conclusion that the system was a waste. Likewise, Phase 3 investment can be scoped accurately only after Gates 3 and 4 have stabilized in Phase 2.

As material for the investment decision, building up the cost per mis-shipment using your own actual figures makes the case persuasive. Reshipment freight, chartered special deliveries, labor spent on emergency response, effort spent on corrective action reports at the customer, and the loss of trust that eventually shows up in commercial terms. In Thailand, customers are often spread across several industrial estates, so even a single emergency delivery costs meaningful travel time and money. Put that build-up next to the investment for each phase and the skeleton of your approval request writes itself.

Three reasons mis-shipment countermeasures pay off especially well in Thailand

The same countermeasure tends to produce better results at a Japanese-owned factory in Thailand than at a plant in Japan, for structural reasons.

The first is the difficulty of securing people and the unpredictability of retention. According to JETRO’s 2024 regional analysis report, 40.4% of Japanese companies in Thailand face a labor shortage. Thailand’s unemployment rate stays persistently in the 1% range, and in practice the gap is filled by migrant workers from Laos, Cambodia, and Myanmar. On a floor that assumes turnover, the Japanese style of quality assurance where a veteran notices the problem by eye simply does not hold. Barcode verification returns the same result regardless of years of experience or native language, so its relative effect in this environment is larger. If you choose terminals whose display language can switch to Thai or Burmese, training time shrinks further.

The second is that throwing labor at the problem keeps getting more expensive. Thailand is moving toward a nationwide flat statutory minimum wage of 400 baht, and the ruling party has pledged 600 baht by 2027. A countermeasure built on adding double-check staff accumulates cost every year as wage levels rise. Investment in a verification system, by contrast, is weighted toward initial cost, and labor hours do not scale with shipping volume. Viewed over a few years, the crossover point arrives sooner than expected.

The third is that the physical burden of each single mis-shipment is heavier. Customers are dispersed across a wide area of industrial estates, and where shipments involve export, you also face corrections to invoices and packing lists and a redo of customs clearance. Some cases simply cannot be resolved the way domestic delivery can, by driving it over the next day, and that one case can affect how your entire account is evaluated. The value of keeping errors inside the factory walls is higher here than at a site in Japan.

Four places implementations get stuck

Knowing where projects typically stumble, before you choose a system, helps you avoid stalling after go-live.

The first is masters that are not in order. Old and new part numbers coexisting for the same component, full-width and half-width characters mixed in notation, pack quantities never registered, obsolete codes never deleted. Load a verification system on top of that and operators start getting errors while working correctly, so the floor invents a procedure for skipping verification. Cleaning up masters is unglamorous, but it is the part of an implementation project whose effort is most often underestimated.

The second is failing to design for exceptions. Emergency shipments, direct shipment to the customer, sample shipments, free-issue material, reshipment of returns. Build only the normal flow and the floor will process every exception outside the system. Anything processed outside makes stock disagree, which creates Gate 2 errors. Think of exceptions not as something to eliminate but as something that needs a path through the system.

The third is the fit between hardware and the physical environment. Wireless not reaching the back of the warehouse or behind metal racking, terminals condensing when carried from an air-conditioned office into a hot and humid warehouse, scanner windows fogging in dusty processes. Build in a step where real equipment is brought in before purchase and tested in the worst conditions on site.

The fourth is the design of KPIs and operating rules. Define what counts as a mis-shipment, record internally caught cases separately from escapes, and review by gate every month. Without that mechanism you cannot show the effect of the system in numbers, and you cannot build the case for the next investment. Start measuring the mis-shipment rate at the same time you go live.

Frequently asked questions

What is the single most effective mis-shipment prevention measure

There is no universal answer. It depends on which gate creates your errors. The fastest route is to classify the past year of mis-shipments by gate and act on the gate with the highest count. In many parts factories, picking at Gate 3 is the largest source, and in that case handheld verification offers the best return. If your history concentrates in Gate 1, mechanizing order entry comes first. As a foundation common to every gate, consolidating item tag and label issuance into a single source is what works.

How much does a shipping verification system cost

Cost varies too widely to state a single figure, because it depends on process scope, terminal count, whether integration with existing systems is needed, and the number of sites. When comparing quotations, list initial cost, meaning terminals, printers, and build, separately from annual running cost, meaning maintenance, label media, and licenses. For implementations in Thailand, using baht-denominated initial and annual costs as the axis, with yen conversion shown only for reference, speeds up internal decision-making. With a phased approach, the Phase 1 investment can be kept limited.

Should we choose handheld terminals or RFID

For a factory with no verification mechanism today, starting with barcode verification using handheld terminals is the sensible move. The investment is smaller and there are fewer constraints on which packaging forms it applies to. RFID reads many tags at once, so it is powerful for pallet-level handovers and pass-through verification, but reading becomes unstable near metal or liquids, and the per-tag cost recurs indefinitely. Rather than picking one exclusively, the realistic configuration is handheld verification as the base, with RFID considered only for processes where bulk reading pays, such as a dispatch gate. Selection criteria are covered in detail in selecting barcodes, QR codes, and RFID for traceability.

How far can the mis-shipment rate realistically be reduced

Automated distribution centers achieve levels around 10 PPM. Warehouses under analog control can sit at around 1000 PPM, a spread of one hundred times. For a site working on improvement, below 100 PPM is a realistic first target, and 100 to 300 PPM is generally cited as normal for e-commerce logistics. That said, these figures assume Gate 3 onward has been locked down with verification. If Gate 1 entry errors remain, a certain number of cases will persist as a floor no matter how much you strengthen verification.

Can we work on mis-shipments while staying on Excel

Partly, yes. Standardizing the format of shipping instructions, cleaning up inconsistent part number notation, and managing pack quantities as a master in one place are all effective while still on Excel. However, Excel cannot give physical items barcodes for machine verification, so you cannot build a mechanism that stops Gate 3 and Gate 5. And when several people update the file simultaneously, version conflicts arise, which tends to increase Gate 1 and Gate 2 errors rather than reduce them. Even if you stay on Excel for now, switching label issuance to come from the system that holds the inventory first will make the eventual migration much easier.

Summary

The first thing to check in mis-shipment prevention is not how strict your inspection is, but which gate is creating the errors. Of the five gates, order entry and shipping instructions, matching stock data to physical items, picking, packing and piece counts, and label application and destination, shipping inspection reliably stops Gate 3 and Gate 5, plus part of Gate 4 if you verify at package level. Errors created at Gates 1 and 2 pass through while barcode verification operates perfectly. Adding inspection staff without understanding this only accumulates cost.

A sound sequence of countermeasures is to consolidate item tag and label issuance to build the Gate 2 foundation, then lock down Gates 3, 4, and 5 with handheld verification, and finally address Gate 1 by mechanizing order entry. In parallel, start measuring the mis-shipment rate in PPM and record internally caught cases separately from cases that escaped to the customer. Set below 100 PPM as the first target and aim from there for the 10 PPM range. Without that sense of where you stand, you cannot explain the effect of the investment to anyone.

At sites in Thailand, workforce mobility, continuously rising wages, and the heavy physical burden of each individual case combine so that systematizing verification tends to pay back faster than it would in Japan. If there is any prospect of being asked to support SSCC and ASN, building a structure that can hold a unique number per package into the design from the start will save you rework later.

Which gate creates your mis-shipments is something you can determine internally by classifying your own history. Still, having examples of where the concentration sat at other sites makes it faster to see the shape of the solution. TOMAS TECH builds production management, inventory management, and traceability systems for Japanese manufacturers across Thailand and ASEAN, and we are happy to take questions purely about separating the gates or sequencing a phased rollout. Even at the stage where you only want to organize your current dispatch operations, feel free to reach us through our contact page.

References

• Calculating the mis-shipment rate in PPM and its benchmark levels (in Japanese) — https://n-apt.com/info/wrong-shipment-rate/
• What PPM means as a logistics quality standard (in Japanese) — https://transcosmos-ecx.jp/blog/fulfillment/35
• Logistics improvement guide 2026 for reducing mis-shipments, with PPM targets (in Japanese) — https://stockcrew.co.jp/insights/article_kpi_miss_ship_rate_2026-05-04
• SSCC, Serial Shipping Container Code — GS1 Japan (in Japanese) — https://www.gs1jp.org/standard/identify/sscc/
• Toward zero mis-shipments through automated shipping verification (in Japanese) — https://www.nsight.jp/blog/shipping-verification-mis-shipment-prevention/
• Causes of mis-shipments and countermeasure examples (in Japanese) — https://tebiki.jp/genba/useful/logi-shipping-mistake/
• Handheld terminal verification system case study at a factory in Thailand, NS Solutions — https://www.smri.asia/en/nss/news/3291
• Labor shortage and minimum wage trends in Thailand — JETRO (in Japanese) — https://www.jetro.go.jp/biz/areareports/special/2024/0303/f5b4d6344434b2a9.html