Gloves are supposedly in stock, yet the line-side shelf is empty. A maintenance filter appears in the storeroom ledger but cannot be found. Lubricant is purchased by the can, used by the litre, and never quite reconciles at month-end. A new inventory screen will not resolve these failures by itself. Low-cost MRO and indirect consumables move frequently, often through several shelves after leaving the central store. The practical question is how to record who took what, from where, in what quantity and for which job, without making every withdrawal an administrative task.
This guide is for manufacturing, maintenance, procurement and IT teams at Thai plants. It covers gloves, filters, lubricants and shop towels; item and location masters; point-of-use issues; two-bin and min-max replenishment; ERP and CMMS integration; RFP requirements; and factory and site acceptance tests. Every number below is an explicitly stated illustration, not a measured result at a real plant.
Define the boundary: consumables, critical spares and production materials
IBM’s February 2026 explanation of the MRO supply chain includes lubricants, protective equipment and cleaning supplies alongside replacement parts. This article focuses on inexpensive items used repeatedly across departments. A dedicated motor that may be needed only once in several years requires a different stock decision: downtime impact, lead time and substitutability dominate its policy. Gloves and towels call for attention to issue frequency, replenishment workload, stockouts at individual use points and unrecorded withdrawals. Applying one inventory-turnover target to both categories invites the wrong decision. Our spare-parts inventory guide explains the critical-spares logic separately.
Production materials are different again. They usually follow a product bill of materials, manufacturing orders, lots, work in process and product costing. Putting every glove into the production BOM creates unnecessary maintenance of that structure. Some filters or chemicals, however, do require batch or use-history controls for quality or safety. Set traceability by item and use case, rather than declaring every indirect item identical. For a broader comparison of system categories, see our factory inventory system comparison.
Set the objective at the shelf, not only in the financial ledger
Management may ask to reduce consumables spending while operators ask to avoid stockouts. Both needs belong in one design. Cutting the central stock figure is no benefit if line-side shelves run empty and procurement must pay for repeated urgent deliveries. Filling every shelf creates redundant holdings, expired stock and hidden inventory. The goal is availability at the point of use with a controlled cost of replenishment and ownership.
Map three quantities separately: the central store balance, the line-side shelf balance and unused items already taken by a worker. If a transfer to the line is treated as final consumption, the ledger decreases even though physical stock remains nearby. If ownership stays in inventory until the item is actually used, someone must record each use. Choose the boundary by value, frequency, loss and safety risk. A mixed policy is legitimate: charge ordinary gloves to a department at replenishment, but issue a special filter against a maintenance work order when installed.
Study a day’s movements before counting every SKU
Follow selected items from purchase order to receiving, storage, transfer, line-side shelf, actual use and disposal. Record who touches the system at each step, where paper is copied later and where the network is weak. Observe a night shift, not just the day shift that normally hosts a software demonstration. If the storekeeper records day issues but night workers take supplies and adjust the sheet the next morning, the major source of variance is outside the demo.
The useful discovery output is an event table. For each receipt, put-away, transfer, replenishment, issue, return, damage, expiry, count and adjustment, specify actor, time, location, unit, device, approval and cost destination. An adjustment needs a reason code and a separate permission. An anonymous change to a balance cannot explain whether stock was consumed, misplaced or counted incorrectly. Decide how corrections are linked to their original transaction.
Build an item master around identity and units
The same glove can be entered under Japanese, English, Thai, a supplier part number and a short nickname. If these become different item codes, the plant cannot see combined holdings or spend. Keep a stable internal item ID, localized display names and supplier references in separate fields. Distinct sizes or materials that are not interchangeable need distinct SKUs. Equivalent approved products can have a documented substitute relationship, but safety items should not be substituted without the relevant site approval.
The minimum record should cover ID, class, names, specification, purchase unit, stock unit, issue unit, conversion, storage condition, expiry and lot requirements, approved substitutes, supplier and lead time. Locations need their own IDs. If oil arrives in cans and is issued in litres, define the volume of a can and the method for measuring a part-used can. If gloves arrive in boxes, keep the number of pieces per box current when a supplier changes. A CSV migration that carries quantities but ignores units turns boxes into pieces or vice versa. Microsoft’s unit-of-measure documentation illustrates why conversions belong in product setup.
The barcode does not fix a unit problem on its own. Its meaning must match the master and the label: does it identify a box, a piece or a particular batch? Control who may change conversions and how historical transactions keep their original meaning after a change.
Before go-live, do not import every row of the old ledger as usable stock. For each item, check whether it is still used, whether the physical item matches the record, and who approved its unit conversion. Gloves with similar names must not be merged if their oil resistance or safety specification differs. If the same filter is registered twice under supplier-specific names, record its approved substitution relationship before consolidating the records. Keep old and new codes, purchasing and stock units, conversion factors, locations, reviewer and review date in the migration map. Do not release unverified items as available stock in bulk. Issues made between the physical count and the data cutover can create a difference, so decide which ledger receives transactions during the transition.
Testing unit conversion requires more than inspecting the item master on a screen. Receive a real box, open part of it, issue some units to a line, and return or transfer the remainder. Check whether the quantities in each screen reconcile to the same base unit. If one terminal defaults to the purchasing unit while another defaults to the stock unit, a discrepancy can arise even when workers follow the displayed steps. After a unit or specification change, confirm that historical quantities and costs can still be reconstructed using the terms valid at the time. When correcting a wrong factor, identify affected transactions and record an approved adjustment instead of silently rewriting their history.
Model locations through the point of use
Consumables often run out at a specific machine or line even when the plant total is positive. Distinguish the central store, building, line, machine area and two physical bins. Give each location an owner and a replenishment source. Match shelf labels to the physical layout and to the hierarchy in the application. A QR label is useful only if operators can find the labelled shelf and the recorded location is current.
Multiple line-side locations create an important trade-off. A total plant balance can suggest a transfer from line B to line A, but the team must see that information and record the move. Counting every low-value use point after each glove withdrawal can become excessive. Some plants should track central stock accurately and control local shelves by replenishment status, while applying finer traceability to selected safety or quality items. Write these exceptions down rather than allowing them to grow informally.
Make the issue transaction fit the act of taking stock
If an operator must return to an office PC, search a catalogue, enter a quantity and select a department for every inexpensive item, transactions will be delayed. Design the shortest feasible path at the use point. A shelf scan can preselect item and location, leaving quantity and purpose. For fixed-quantity replenishment, an empty box can trigger the request. For supplies charged to departments, the storeroom may issue a batch to the department. In each case, one real-world movement should correspond to a clear ledger event.
Issue to a work order when maintenance needs item cost by asset or job. Microsoft describes consumption registration on work orders, mobile recording of consumed items and an asset BOM that may include lubricants as well as spare parts. These are documented examples of one platform, not features guaranteed in every candidate system. Ask vendors to demonstrate both ordinary departmental issue and issue against a maintenance work order. Include emergency work that starts before a work-order number exists: who records it provisionally and who confirms it later?
Returns and mistakes also matter. A sealed unused filter may be returned against the original issue; an opened chemical may be ineligible for return under site rules. Test cancellation, wrong scans, double taps, interrupted connections and retransmission. A unique transaction ID should stop a retry from reducing stock twice. The acceptance test should intentionally send the same issue twice and verify that the balance changes once.
Use two-bin or min-max replenishment with named ownership
In a two-bin method, a working bin and a reserve bin hold the same consumable. Emptying the working bin produces a replenishment signal, and the reserve covers demand during replenishment. This can suit regularly used gloves or towels. The application alone cannot make it work: decide who collects an empty-bin card, what happens at night or on a weekend, and when the reserve is considered opened. A visual card, QR scan and weight sensor are alternatives; compare workload and the fallback when the device fails. A sensor is not a default requirement for every SKU.
Under min-max control, each location has a lower and upper level; when available stock reaches the lower level, replenish toward the upper level. Align the upper level with shelf capacity and purchase or transfer units. Decide when ownership moves from store to line. Review values when demand, supplier performance or working days change. Initial levels are hypotheses to refine against observed stockouts, excess and replenishment frequency. If usage suddenly increases, investigate the purpose and missing issue records before simply raising safety stock.
For illustration, assume one line consumes two boxes of gloves per working day, replenishment takes three working days, and the site wants one extra day’s allowance. A lower-level candidate is 2 × 3 + 2 × 1 = 8 boxes. If the provisional upper level is 14 and five boxes remain when replenishing, move nine boxes. The calculation assumes stable average use, a reliable three-day replenishment time and one consistent “box” unit. Do not copy the eight-box number to irregular or long-lead-time items. Define whether available stock includes open purchase orders or stock in transfer and how to round to pack size.
Replenishment rules must cover the case where the reserve bin is also empty. If demand rises and a line borrows from another line, define who authorizes the transfer and when quantities at both locations change. Check that emergency purchases delivered directly to the line are not also counted as receipts into the central storeroom. If no replenishment owner is present on the night shift, document usable reserve stock, the contact route, and the next-morning reconciliation. Sensors and automated orders do not eliminate exceptions such as items moved to another shelf or an empty-bin card that was never returned. During a weekly walk-through, compare shelf labels, physical items and system replenishment status. Assign an owner to distinguish and correct a wrong setting, a demand change, a missed transaction or a supply delay.
With vendor-managed or consignment stock, distinguish items physically on the shelf from stock legally owned by the factory. Contract terms may place ordering, delivery, acceptance, transfer of title and expense recognition at different points, changing which transaction ERP receives for the same apparent replenishment. In the RFP, test whether the system can show owner, location and available quantity separately rather than reporting only a combined balance for consigned and ordinary items.
Draw the ownership boundary between ERP, CMMS and the shop-floor app
A stand-alone tool can leave receiving, payables, department expenses and maintenance costs in separate ledgers. One approach is to make ERP the master for items, suppliers, purchase orders and accounting, while a shop-floor app handles local issues and replenishment. If the existing ERP already supports these flows, extending the device workflow may be simpler than creating a second stock balance. In the RFP, identify the system of record for item master, physical balance, valuation, approved purchase order and maintenance work order. “API integration available” does not answer who owns a conflicting update.
For CMMS or EAM integration, specify where work orders originate, what statuses permit issue, what cancellation does, and how asset IDs and cost centres travel. Microsoft’s mobile work-order material is a useful example of workers registering time and items against assigned jobs. A vendor demonstration should begin with the work order, scan the item, confirm quantity and end at the resulting ERP cost line. A screen mock-up alone cannot prove that units and financial dimensions survive the interface.
Real-time synchronization is not automatically best. Where the network fails, confirm offline storage, idempotent resend and an explicit maximum age for balances used in replenishment alerts. For daily batch exchange, show pending transactions in the operator’s view and prevent double counting in reorder proposals. Define who checks the retry queue, how errors are reported and by when a mismatch must be reconciled.
Decide cost-centre rules before configuration
The level at which consumables are charged changes the operator’s input burden. Charging every glove to a work order may appear precise but may increase omissions. Finance and operations should choose the required granularity by item class: ordinary gloves to a department, special filters to a maintenance order, lubricant to a group of assets, for example. Do not create fictitious work orders merely to satisfy a mandatory field.
At Thai plants, purchase and accounting currencies may differ. Correct boxes on an operator screen do not prove correct valuation if conversions or period-end accounting rules are wrong. Trace a price change, return, disposal and stock adjustment through the accounts and cost centres during testing. Accounting policy belongs with the company’s finance team; a vendor default must not silently decide it.
Compare products with the same scenarios
A feature checklist may show ticks beside “barcode,” “reorder point” and “ERP interface” without proving that a usable process exists. Give each vendor identical sample data and scenarios: receive boxes and issue pieces; transfer stock to another use point; issue offline on a night shift and synchronize once; charge a filter to a work order and reverse a cancellation; trigger replenishment from an empty bin and round to supplier pack size; record expiry separately from consumption. Follow each scenario from screen operation to balance and cost entry.
Record whether each step is standard functionality, configuration, customization or manual work. Compare operator actions, exception handling, master maintenance, permissions, audit history, interfaces and total ownership cost. A ten-screen issue flow will be abandoned at the shelf. A one-scan flow that loses the cost centre creates work downstream. Include labels, devices, connectivity, cleansing, training, integration and support in the estimate, not just the initial software price.
Put testable boundaries in the RFP
State which factories, buildings, lines, use points and item classes are in scope. SKU count alone is not enough. Daily transaction volume, night-shift usage, weak-signal areas, unlabeled items, languages and exceptional returns all affect effort. Specify whether procurement approval and counting are included, or only point-of-use issue.
List required, desired and out-of-scope functions for receiving, transfer, issue, return, disposal, adjustment, replenishment, counting and history. List data fields including localized names, units, cost centres, lot and expiry, supplier references and substitutes. Set requirements for separated permissions, logs, backup, recovery, device failure and retention. For each interface, name direction, frequency, identifiers, retry behavior, reconciliation and owner. Add a sentence explaining what an observer must see for each requirement to pass a test.
Require separate estimates for data migration, local training, Thai-language UI, device setup, ERP/CMMS integration, FAT, SAT and post-launch support. State expected user counts to compare monthly and additional-user fees. “Can be developed later” is not delivered scope. Agree how change requests and failed acceptance tests will be handled before signing.
Give every vendor the same sample data and exceptions for the RFP comparison. Fix conditions such as gloves purchased by the box and issued by the pair, a night-shift network outage, an emergency issue without a work order, and an opened item that cannot be returned. Classify each response as demonstrated standard function, configuration, quoted custom development, or a remaining manual step. Do not treat an item that was only described by sales staff as a passed demonstration. Similar quotations may include different numbers of terminals, label printing, master-data migration, local training, language support and post-launch help. Show first-year and later costs separately, together with prices for added lines and users.
Define acceptance before signing the contract in terms of operational outcomes. Examples are a single stock deduction after two sends of the same offline issue, a physical balance that reconciles after box-to-piece conversion, and no cost left against a canceled work order. Retain transaction ID, timestamp, operator, before and after quantity, and the receiving system’s response as evidence. Manage each defect with its shop-floor impact, workaround, owner and retest date. A Thai-language interface alone is not proof of usability; ask actual workers to check item search terms and unit displays.
Test failures as well as the normal flow in FAT and SAT
Use factory acceptance testing (FAT) to verify configuration and interfaces before deployment; use site acceptance testing (SAT) with actual shelves, devices, workers and network conditions. In FAT, test boxes versus pieces, a glove bin, a filter issued to a work order and a part-used oil can. Add returns, cancellation, duplicate transmission and an ERP rejection. Record expected balances and costs before executing each test and compare actual results afterward.
In SAT, check whether labels are legible, devices can be used with protective gloves, connections work at remote shelves, night permissions are correct and pending transactions survive shift handover. Include a device-failure fallback such as temporary paper capture followed by controlled entry. Test with operators and replenishment staff as well as administrators. Passing means that movements are recorded, variances can be explained and staff can handle exceptions. Log defects with severity, workaround, owner and correction date.
Measure KPIs with a stable denominator
An inventory-value KPI alone can improve on paper when stock is simply transferred to line expense. Monitor stockouts at use points, urgent orders, replenishment time, physical versus book variance, expiry and disposal, issue-recording completeness and total holdings together. Define an urgent-order rate as urgent consumables orders divided by all consumables orders, if that is the chosen measure. A stockout rate might use item × use point × working day as its observation unit. Choose and keep the denominator before claiming improvement.
Compare before and after periods with similar working days and production conditions. A new line or change in volume can move total spend without any process change. Show production volume, operating hours, headcount or work-order counts alongside cost where relevant. Separate price and currency effects from quantity. Set targets from the site’s baseline and risk tolerance; generic percentage savings are not evidence of a particular plant’s result.
An illustrative calculation, with its assumptions visible
Suppose a fictional plant has 12 urgent consumables purchases per month and assumes THB 1,500 of extra freight and valued administration time for each. If the same definition yields four urgent purchases after process changes, eight avoided events × THB 1,500 = THB 12,000 per month. Suppose replenishment checking falls from 40 to 20 minutes a day over 22 working days, with time valued at THB 300 an hour. Then 20 ÷ 60 × 22 × 300 = THB 2,200 per month. The combined candidate benefit is THB 14,200 per month.
Subtract subscriptions, device and label maintenance, master-data work, variance investigation and training. A simple initial-cost ÷ monthly-benefit payback ignores timing and durability of costs and benefits. These figures are explanatory assumptions, not a case study. In a real RFP, use measured urgent-order costs, a stated method for timing tasks and sensitivity analysis for a smaller improvement. Include avoided production stoppage only when its causal link and actual duration have been verified, and avoid double counting.
Pilot representative processes before expanding
Select items that expose different workflows: common gloves or towels, a filter charged to a work order and oil requiring unit conversion. Include use points with different connectivity and staffing. Clean the master, label the shelves, assign transaction owners and train users before running the pilot. Define a limited parallel-run period and who investigates differences between old and new ledgers. An indefinite parallel run only doubles data entry.
Expand against observed criteria: night-shift issues are traceable; replenishment requests are completed; ERP balance differences can be explained; and operators know what to do with exceptions. Correct duplicate items and units before adding another line. Name the person who reviews min-max levels after go-live. A functioning application will still drift if shelf ownership and settings are left unchanged when the process changes.
Conclusion
A factory consumables inventory system should describe movements and cost ownership through the point of use, not simply display a central stock figure. Clean item, unit and location masters; select low-friction issue and replenishment rules; identify the authoritative ERP and CMMS records; and test real shop-floor scenarios in the RFP, FAT and SAT. Explicitly separate ordinary consumables from critical spares and production materials.
TOMAS TECH can help Thai manufacturing sites map line-side shelves and existing ERP/maintenance workflows into transaction rules, master fields and vendor test scenarios. If you want to start by mapping how gloves, filters and oil are issued and replenished today, contact our team.
Frequently asked questions
Must every consumable be recorded as individual pieces?
No. Choose granularity by value, frequency, operator workload and quality or safety requirements. Ordinary gloves might move by box to a line; a special filter may need an item count and work-order reference. Maintain purchase-to-stock unit conversions in the master.
Does a two-bin system remove the need for software?
A small operation with a stable replenishment owner may manage with physical cards. Multiple lines, spend visibility and ERP-linked purchasing require a way to capture and consolidate the empty-bin signal. Physical replenishment rules and digital records solve different parts of the process.
Does VMI eliminate internal recording?
Even with vendor-managed inventory, the contract and site process must define when items are used, when expense is recognized, and who pays for damage or expiry. Separate supplier replenishment responsibility from internal point-of-use control.
What if there is no work order?
Charge general consumables to a department, line or cost centre under an agreed rule. Link items directly used in maintenance to real work orders. Fictitious orders may satisfy a screen requirement while destroying the meaning of the history.
When should benefits be judged?
Measure a baseline for stockouts, urgent orders, replenishment time and variance first. Compare the same definitions after launch and account for changes in production volume, working days and new lines. Total spending alone is insufficient.
