Inbound and Outbound Management System, Three Timing Designs

One of the most common questions we hear from factories in Thailand is why book inventory still fails to match physical stock after an inbound and outbound management system has been installed. In most cases the software is not the problem. The gap sits in the time lag between the moment goods physically move and the moment the data is updated, in other words in how recording timing is designed. This article sorts recording timing into three patterns and uses an original model-factory calculation to show where each one breaks even.

The cause is recording timing, not the software

Warehouse inventory data is built from individual movement events, namely receiving, issuing to production, returning from production, shipping, and rack-to-rack transfers. Every time one event is recorded, book inventory advances one step.

What gets overlooked is that the time an event happens and the time it reaches the database are not necessarily the same. If a pallet moved from the receiving dock to the rack at 9 a.m. but the fact was keyed in at 11 a.m. the following day, the system will insist for 26 hours that the pallet is still on the dock.

During those hours a buyer sets order quantities from that screen, production control allocates material, and the shipping team issues picking instructions. The judgement may be sound and the data entry may be flawless, but if the underlying data is stale the outcome will be wrong. Factories feel that things drift despite having a system because this time difference is never counted as an error source.

When you look into the causes and countermeasures for inventory discrepancy, separate them into three groups first.

• Movements that were never recorded, such as lost paper slips or missed scans
• Movements recorded with wrong content, such as transcription errors or part number mix-ups
• Movements recorded correctly but reflected late, in other words time lag

Only the third one refuses to shrink when you push operators to be more careful. It responds to operational design alone. How to set the level of stock you hold is covered separately in our guide to optimal inventory level management, but whether you can hold that level depends on the accuracy of the recording timing discussed here.

Three patterns of recording timing

Recording timing as observed on real shop floors falls into roughly three patterns. None of them is inherently right. The correct choice depends on scale and on the state of your supplier base.

Pattern A, batch recording after the fact

The floor runs on paper slips and whiteboard notes, and an office clerk keys everything into the system daily or weekly. Additional investment is close to zero and the physical workflow is untouched. In exchange, a lag of half a day to several days opens up between the record and the actual move, and during that window book inventory is only a reference figure. Across a weekend, up to three days of movements can pile up unposted.

Pattern B, real-time scan recording

Every receipt, shipment, and rack transfer is captured with a handheld terminal or tablet reading a barcode or QR code, and inventory updates on the spot. The lag essentially disappears and real-time inventory management becomes possible. The catch is that a new action, scanning, has to be built into the physical workflow, and missed scans during busy hours become a fresh error source. Terminal selection and rollout are covered in our article on handheld terminal deployment.

Pattern C, pre-registration with confirmation by matching

Purchase order data or an ASN (advance shipping notice) is loaded into the system in advance, and at receiving a scan simply confirms that the plan and the physical goods agree. Because nobody keys in quantities, the floor workload is the lightest of the three, and any gap against the order, meaning short quantity, wrong item, or damage, is caught on the spot. The detection mechanism is continuous with the design of a shipping and receiving inspection system. Its weakness is the prerequisite, since it only works when purchase order accuracy and supplier cooperation are both in place.

Placed side by side, the three patterns look like this.

Patterns B and C are alike in that both all but eliminate the time lag. Where they differ is in what error is left behind. Pattern B still carries 1.2% of missed scans, while Pattern C falls to 0.3% because the manual quantity-entry step disappears entirely, removing one source of error, and matching against plan lines also makes any omission visible. Detection speed differs across all three. An error under Pattern A sleeps until stocktaking, an error under Pattern B is caught by same-day reconciliation, and an error under Pattern C surfaces automatically on the next morning’s list of unmatched plan lines. No pattern drives error to zero. What you should choose is the pattern whose time to visibility fits inside your own decision cycle.

Model factory S, the basis of the calculation

To turn the argument into money, we set up a model factory with the following conditions. It is built specifically for this article and does not reuse the models from our other posts.

The hourly figure comes from a monthly warehouse labour cost of 22,500 THB including allowances and statutory benefits, annualised to 270,000 THB and divided by 2,000 working hours per year. The supervisor rate of 270 THB is derived the same way. Every calculation below uses these two rates.

How much inventory does the time lag hide

Dividing 300 transactions per day by 8 hours gives 37.5 movements per hour. Multiplying that by each pattern’s average lag shows how much movement is permanently sitting unposted.

The conversion assumes an average inventory value of 3,200 THB moved per transaction. Under Pattern A, 2,160,000 THB of stock, roughly 7.7% of average inventory, is permanently in a state where the system says it exists but it has in fact already moved. Under Patterns B and C this falls to 32,000 THB, about 0.11%. The ratio between them is 67.5 times.

The same gap drives the workload of physical counting. Tracing a discrepancy across 675 events is a completely different job from tracing it across 10. How to run the count itself is covered in our guide to physical inventory efficiency, but shrinking the population you have to trace is the job of recording timing.

Annual cost of the three patterns

At 300 transactions per day, here is what each recording design costs per year, built up item by item. The assumed error rates and unit costs are as follows.

• Recording takes 24 seconds under Pattern A, 12 under B, and 8 under C. At 135 THB per hour that is 0.90 THB, 0.45 THB, and 0.30 THB per event
• Recording error rate is 1.2% for Pattern A (0.4% lost slips plus 0.8% transcription errors), 1.2% for Pattern B (missed scans), and 0.3% for Pattern C
• Investigating one discrepancy takes 60 minutes under A, 20 under B, and 12 under C, or 135 THB, 45 THB, and 27 THB
• One line stoppage caused by a stockout costs 25,000 THB, being 20,000 THB of lost gross profit over two hours plus 5,000 THB of recovery labour
• Correcting one mis-shipment costs 3,150 THB, being 1,800 THB of re-delivery freight, 540 THB for four hours of rework on the floor, and 810 THB for three hours of supervisor time with the customer. Against 15,000 shipments a year the mis-shipment rate is 0.4% for Pattern A, 0.12% for Pattern B and 0.08% for Pattern C, which works out to 60, 18 and 12 cases
• Up-front investment is amortised over five years. Pattern B costs 300,000 THB for five terminals, wireless coverage, and training. Pattern C adds the purchasing system interface and ASN rollout for a total of 800,000 THB
• Maintenance and terminal renewal runs at 30,000 THB a year for Pattern B, and at 54,000 THB for Pattern C, being 30,000 THB of terminal maintenance plus 24,000 THB for the interface

Annual line stoppages are set at 15 for Pattern A, 6 for Pattern B, and 3 for Pattern C. The 15 under Pattern A break down into 9 caused by recording lag and 6 caused by demand swings and supplier delays. Under Pattern B real-time stock cuts the first group to 2, and faster emergency sourcing decisions bring the second group down to 4. Under Pattern C the ASN gives advance warning of supplier delays, so the two groups become 1 and 2. This is where stockout prevention works qualitatively differently between B and C.

All figures are in THB. Moving from Pattern A to Pattern B saves 382,050 THB per year, and moving to Pattern C saves 397,625 THB. Those savings, however, already have the amortised investment deducted as a cost, so feeding them straight into a payback calculation would count the investment twice. Adding the amortisation back gives a cash saving of 442,050 THB for Pattern B and 557,625 THB for Pattern C, and dividing the up-front investment by those figures puts payback at about 8 months for Pattern B and about 1.4 years for Pattern C.

The striking result is that Patterns B and C are only 15,575 THB apart. At 300 transactions per day, two designs whose investment differs by nearly 2.7 times end up almost level. The break-even point has already been passed, but the gap is small enough that starting with Pattern B while supplier integration is not yet in place, then moving to Pattern C in stages once it is, is also a reasonable call.

Managing missed scans

Moving to Pattern B does not reduce how often a recording error occurs; our calculation puts the missed scan rate at 1.2%, the same as Pattern A’s recording error rate. The annual cost still falls by more than half, and the reason is that the misses are caught the same day, before they grow into stockouts or shipping mistakes.

Three mechanisms make same-day detection possible. The first is an operating rule that drives the receiving dock and the shipping lane to zero stock at the end of each day, so any physical item still sitting there is an obvious missed scan. The second is a daily list of scan counts by terminal, so an unusually quiet terminal gets checked the next morning. The third is an automatic extract of SKUs whose day-over-day movement disagrees with the theoretical figure.

None of these requires custom development, only operating rules combined with existing reports. The 20 minutes per discrepancy assumed for Pattern B presumes that these mechanisms are running. Without them, investigation reverts to the 60 minutes of Pattern A and the discrepancy line item swells to 121,500 THB instead of 40,500 THB. The annual total for Pattern B then becomes 451,950 THB and the gap against Pattern A narrows from 382,050 THB to 301,050 THB. The investment is not wasted, but 81,000 THB a year of the benefit is left on the table. That is why handing out terminals cannot be the end of the project.

Where each pattern breaks even by daily volume

Splitting the cost items into variable costs that scale with volume and fixed costs that do not gives the following. Five terminals are assumed to cover up to 500 transactions per day, so fixed cost is treated as flat within that range.

The break-even points are where these three lines cross. Converted back at 250 operating days, Patterns A and B swap at about 57 transactions per day, A and C at about 105, and B and C at about 267. Totalling by volume gives this picture.

All figures are in THB. Put into words, paper stops paying its way somewhere past 57 transactions a day, and involving your suppliers only starts to earn its keep somewhere past 267. Conversely, a plant handling 40 a day that aims straight for Pattern C will pay an extra 132,450 THB a year. Note that this table assumes line stoppages and mis-shipments rise and fall in proportion to transaction volume as well. Note too that a fixed cost covering five terminals is excessive for a plant at 40 a day, and a single-terminal setup would make the fixed cost of Pattern B much smaller. Treat the crossover at 57 per day as a conservative figure that assumes five terminals.

These numbers rest on the assumptions of model factory S. Error rates and stoppage costs move with industry and product value, but the underlying structure does not change, namely that the earlier pattern stays cheaper until volume crosses a threshold. Cutting excess inventory and preventing stockouts both start with working out which of the three lines your own plant is sitting on. For choosing a product type see our factory inventory system comparison, and for the wider cost picture see our breakdown of WMS implementation cost.

Design location control and FIFO at the same time

Once recording timing is tightened, the next lever is the granularity of location control, meaning how precisely you say where an item sits. Finer warehouse locations cut search time, but every rack transfer now has to be recorded, so the number of recorded events rises. Location granularity therefore feeds straight back into the variable cost above.

Introducing bin-level location control while staying on Pattern A only raises the number of records without shortening the lag, so discrepancies get worse rather than better. Granularity should be tightened only after recording has become immediate.

A FIFO management system follows the same order. Any mechanism that controls issue sequence by receipt date or lot number assumes that receipts are recorded immediately. Under Pattern A an issue can happen before its receipt is posted, so the sequence the system dictates will not match the physical arrangement on the rack. If FIFO matters to you, move the inbound side to Pattern B or C first.

On the outbound side, picking instructions based on stale inventory data are a breeding ground for mis-shipments. The full countermeasure set is in our article on preventing shipping errors.

Conditions that add up in Thailand

Designing warehouse inbound and outbound management in Thailand is not quite the same exercise as in Japan.

First, customs. Thailand’s e-Customs system has handled electronic declarations since 2007, and commercial importers and exporters have long operated under a paperless licence for online import and export filing. The groundwork for receiving advance electronic data on imported parts is already in place, which makes the data source that Pattern C depends on easier to secure.

Second, logistics cost. The National Economic and Social Development Council (NESDC) is reported to have set a target in the 13th National Economic and Social Development Plan of lowering logistics cost as a share of GDP from 13.8% in 2021 to below 11% by 2027. Streamlining import and export clearance procedures is among the priority areas, which puts pressure on shippers to improve turns and cut dwell time.

Third, the state of automation. Some market studies indicate that around 16% of new warehouse facilities in Thailand have begun adopting automation, real-time inventory management, or a WMS. This is not a high-precision primary statistic and should be treated as indicative, but read the other way it suggests that most warehouses still sit close to Pattern A. Thailand’s freight and logistics market is forecast at 53.38 billion US dollars in 2025 and 56.56 billion US dollars in 2026, so the market itself keeps expanding.

Fourth, workforce turnover. Warehouse staff change over faster on Thai shop floors than in Japan, and paper-slip operations tend to depend on tacit knowledge held by individuals. When the procedure is embedded in the terminal, as with Patterns B and C, the handover burden drops visibly.

What the Monodzukuri White Paper 2025 says about scale

The situation in Japan is instructive. According to reporting by MONOist on the Monodzukuri White Paper 2025, the areas where manufacturers apply digital technology are clerical work at 43.9%, production control at 43.7%, and manufacturing at 39.9%.

The same reporting shows DX adoption in the manufacturing domain at 67.9% among companies with 301 or more employees against 31.5% among those with 50 or fewer, a gap of 36.4 points. AI use on the shop floor is put at 12.2%, but only among companies that have already digitised their manufacturing processes, suggesting that most companies are still at the recording and visibility stage.

The pattern that adoption falls with company size is consistent with our own calculation. At low transaction volumes Pattern A is the rational choice and there is no need to force in a more sophisticated mechanism. The real problem is a plant that has crossed the threshold and stayed on Pattern A anyway.

A 90-day path away from Pattern A

Switching every item to Pattern B or C at once throws the floor into confusion and drives missed records up. Take it in stages.

Starting with measurement is the point. If you make an investment decision without having measured your current average lag, you have no way to verify the effect afterwards. Plants that skip the first row of that table end up unable to say anything stronger than that things feel better than before.

Three common failures

The first failure is treating the project as finished once terminals have been handed out. Without a weekly review of the missed scan rate, you are back to Pattern A conditions within six months.

The second is subdividing locations without changing recording timing. As noted above, this only increases the number of records and makes discrepancies worse.

The third is designing around Pattern C before suppliers are ready. If 30% of suppliers cannot conform to the ASN format, receiving turns into a dual operation of Patterns C and B, and the judgement load on the floor actually rises. When you move toward Pattern C, work down from the suppliers that dominate by value or by volume and confirm what each can support before you design. Model factory S assumes that the top 5 suppliers account for about 70% of its 90 daily receipts.

What all three have in common is that recording timing advances as nobody’s responsibility. Purchasing watches order data, production control watches allocation, and the warehouse watches physical goods, but the time axis that connects the three is a blank space. Naming one person to report average time lag every month, right at project kick-off, fills much of that blank.

Frequently asked questions

What is an inbound and outbound management system

It is a mechanism that records inventory movement events such as receipts, shipments, issues between processes, and rack transfers, keeping book inventory current. It only works once the design covers not just what to record but when to record it.

How much does an inbound and outbound management system cost

Under the model factory S assumptions, real-time scanning (Pattern B) costs 300,000 THB up front for five terminals, wireless coverage, and training, plus 30,000 THB a year in maintenance. Pre-registration with matching (Pattern C) adds purchasing system integration and ASN rollout for 800,000 THB up front. Figures move with transaction volume and item count, so treat them as indicative.

Where should we start on inventory discrepancy causes and countermeasures

First count your discrepancies split into three groups, namely movements never recorded, movements recorded with wrong content, and movements reflected late. If the third group dominates, training will not fix it. Changing the recording timing design is the only countermeasure.

Does a small plant need real-time inventory management

In our calculation, a plant handling fewer than about 57 transactions a day spends less per year on batch recording after the fact. Measure your volume before deciding.

Will a stockout prevention system reduce inventory

Stockout prevention and excess inventory reduction are two sides of one coin. When book inventory cannot be trusted, the only defence is to carry thicker safety stock, so tightening recording timing ends up compressing inventory. In model factory S, moving from Pattern A to Pattern C cuts line stoppages from 15 a year to 3.

Summary

Inbound and outbound management breaks down not because a system is missing but because of the time lag between the moment goods move and the moment data moves. Sorting recording timing into batch after the fact, real-time scanning, and pre-registration with matching, and running the numbers for model factory S, the optimum shifts from Pattern A to Pattern B at about 57 transactions a day and from Pattern B to Pattern C at about 267. For factory S at 300 a day, annual cost falls from 753,000 THB to 355,375 THB.

The first move is measurement, not investment. One week of measuring your own transaction counts and average time lag will almost entirely determine which pattern you belong on.

If you would like to see this calculation applied to your own transaction volumes, or to review warehouse operations and production control systems in Thailand, please get in touch through our contact page. We are happy to start from measurement on your own floor.

References

• Reporting on the Monodzukuri White Paper 2025, MONOist, confirmed as of August 2026
• National Economic and Social Development Council, logistics cost target in the 13th National Economic and Social Development Plan, as reported by industry analyst publications
• Share of new warehouse facilities in Thailand adopting automation, indicative figure from secondary reporting of a Mobility Foresights market report
• Forecast size of the Thai freight and logistics market, indicative figure from secondary reporting of market research publications
• Thailand’s e-Customs electronic declaration system (in operation since 2007, paperless-licence import and export filing)