GenFrontier homeScope a project
Menu
Reader questionWhat item master data (dimensions, weights, pack hierarchy, unit-of-measure conversions) must be correct before a WMS or integration goes live, and how do we verify it?

Item master data to fix before go-live: dimensions, weights, units

Which item master data a WMS or integration needs before go-live: dimensions, weights, pack hierarchy and unit conversions, and how to verify each one.

field-notes / integrations / item-master-data-readiness
A stack of five metal pans nested inside one another from largest to smallest, the one in the middle printed red

Before a WMS or integration goes live, the item master data for every active item needs four things right: a pack hierarchy (each, inner, case, pallet) with an identifier and a fixed quantity at every level; length, width, height and gross weight at every level, measured one consistent way; unit-of-measure conversions that multiply cleanly between the units you buy, stock and sell in; and the control flags processes read, such as lot, serial and expiry. Verify it three ways: arithmetic checks across the whole file, a stratified sample measured against the GS1 measurement method and tolerances, and tests of each consuming process with real items. Record where every value came from, so that a failed check leads to the right fix.

Item master data is the set of attributes that describe an item independently of any transaction: its identifiers, units, pack levels, measurements and handling rules. This field note lists those attributes by the process that reads them, explains how to measure and convert them, and gives a sampling audit you can run in the weeks before cutover. It does not decide which system owns each field; that is covered in ERP or WMS as system of record. Loading opening balances is covered in moving inventory into a new WMS at cutover.

Why item master errors surface on go-live day

Item data errors stay hidden in an ERP because people work around them, and they surface in a WMS because software acts on them. A buyer knows that a “case” of one item holds 20 and not 24. A putaway algorithm does not know, and it trusts the number in the field.

In an ERP-only world, most item attributes are read by people. Purchasing reads the case quantity on a PO, a receiver counts what arrived and fixes the quantity by hand, and a shipping clerk weighs the carton on a scale before booking it. Wrong master data costs a little time at each step, and nobody sees it as a data problem.

A WMS, a cartonization engine, a rating engine and an EDI translator read the same fields without a person in between. A case quantity of 24 when the real case holds 20 becomes a receipt that is always 4 short per case. A case height entered in centimeters but read as inches becomes a location that is never big enough. Each of these looks like a software defect in the first week, and each traces back to the item record.

That is why item master work belongs in the project plan from the start, not in the last two weeks. In our engagements we start with requirements, and the item fields each process needs are part of them.

Required item fields, listed by the process that reads them

The simplest way to decide which fields must be right is to start from the processes, not from the item record. For each process going live, list the fields it reads and what fails if a field is wrong. Fields that no live process reads can be cleaned later.

Process Fields it reads What fails when a field is wrong
Receiving against an ASN Item identifier and GTIN at each pack level, unit codes, case quantity, lot, expiry and serial control flags Received quantities do not match the ASN; lot or expiry is not captured when it is needed later
Putaway and slotting Dimensions and weight of the case and pallet, stacking limits, storage temperature or zone Suggested locations are too small or overweight; freight lands in the wrong zone
Replenishment Minimum and maximum in the stock unit, case quantity, pick-face capacity Pick faces overflow, or replenishment moves partial cases
Picking and packing Each and inner dimensions and weight, ship-alone flag, case quantity The packing engine picks the wrong carton; orders ship in more boxes than needed
Label printing GTIN at each level, lot, expiry or serial values, pack quantity Labels print data the customer’s scanner rejects
Outbound ASN Pack hierarchy, quantity per carton and pallet, GTIN per level The ASN does not match the cartons on the truck
Parcel rating Package dimensions and weight Quotes and invoices disagree; dimensional charges arrive after the fact
LTL rating Handling-unit dimensions and weight, freight classification data Reclassification and reweigh corrections on freight bills
Order entry and invoicing Sell units, price unit, conversion to stock unit Price per case is applied per each, or the reverse
Cycle counting Count unit and its conversion to the stock unit A count of cases is posted as a count of eaches
Rotation and shelf life Expiry control, shelf-life days, minimum remaining life to ship Expired or short-dated stock is picked first

Two observations from this table guide the cleanup. First, the pack hierarchy and the conversions appear in almost every row, so they are the highest-priority fields. Second, dimensions and weight matter at the level the process handles: a putaway rule needs case and pallet measurements, a packing engine needs each measurements, and LTL rating needs the measurements of the whole handling unit, pallet base included.

For the outbound ASN in particular, the pack hierarchy in the item master is the raw material for the 856’s hierarchical loops. How that structure is built is covered in building an 856 ASN that matches the truck.

Control flags to settle before the first receipt

Some item attributes are not measurements but decisions: whether an item is lot-controlled, expiry-controlled or serial-controlled, which storage zone or temperature it needs, and whether it may ship alone in its own carton. Settle these before the first receipt in the new system, because they decide what the receiver must capture at the dock.

A flag turned on after stock exists leaves the existing stock without the data the flag requires. Stock received without lot numbers cannot be traced by lot later, and stock received without expiry dates cannot be rotated by expiry. Check how your WMS handles a flag change on an item with stock before you plan to rely on one. The safer path is to agree each flag item by item with the people who receive and ship the product, and to treat a later change as a controlled event with its own steps.

The pack hierarchy: levels, quantities and identifiers

A pack hierarchy is the list of packaging levels an item is handled in, from the smallest unit up to the pallet, with the quantity of the level below contained in each. Every conversion and every case-level measurement hangs from it, so it comes first.

A typical hierarchy has four levels:

  1. Each (base unit). The smallest unit you buy, store or sell. For a retail product, the consumer unit.
  2. Inner pack. An optional level, such as a sleeve of 6 inside a case.
  3. Case. The shipping carton, holding a fixed number of eaches or inner packs.
  4. Pallet. A fixed configuration of cases, described as cases per layer (often called the tie) and layers per pallet (the high).

GS1’s GTIN rules tie identifiers to these levels. The pack/case quantity rule says “A change to the number of trade items in a case or a change to the quantity of cases in a pre-defined pallet configuration requires assignment of a new GTIN,” and that “A unique GTIN is assigned at every existing level of the packaging hierarchy above the retail consumer trade item/base unit level.” Its own example is a case that changes from 9 trade items to 12.

That rule has a direct consequence for your item master. A case of 12 and a case of 9 of the same product are different trade items with different GTINs, so they cannot share one “case” row with one quantity. If a supplier changes its case count, the item master needs a new pack level or a new supplier-specific conversion, not an edit of the old number.

Pack hierarchy and conversions for a fictional item Each EX-20118 3.0 x 3.0 x 9.5 in 0.9 lb Case 12 each 12.4 x 9.3 x 9.9 in 11.6 lb Pallet 12 per layer x 5 48 x 40 x 55.0 in 741 lb with base › › Conversions to the stock unit (each) 1 case = 12 each 1 pallet = 60 cases = 720 each Arithmetic checks that must pass Case weight 11.6 lb is at least 12 x 0.9 = 10.8 lb (carton adds 0.8 lb) Pallet height 55.0 in = 5 layers x 9.9 in + 5.5 in pallet base Pallet weight 741 lb = 60 x 11.6 lb + 45 lb pallet base
Fictional example data for illustration. The checks in the red box are the cross-level rules described in this article.

The arithmetic in that figure is the first verification you can run on the whole item file, with no tape measure. If a case weighs less than its contents, or a pallet’s height does not match its layers, at least one number is wrong.

Unit-of-measure conversion design: purchase, stock and sell

A unit-of-measure conversion states how many of one unit make another, for one item. Good conversion design starts with one stock unit per item and expresses every other unit as an exact multiple of it.

Oracle’s PeopleSoft documentation gives a clear statement of the principle. In its description of the UOM conversion hierarchy, it says “The item standard UOM serves as the smallest common denominator for UOM conversion,” that it “should be the smallest unit in which the item is transacted,” and, as an example, “do not assign a standard UOM of CASE to an item that you regularly order in the smaller UOM of BOX or EACH.” The same page separates three levels of conversion: general units and their rates, an inventory item’s own units, and an item-supplier unit that “represents the UOM in which the item is purchased from a specific supplier.”

Most ERPs have some version of those three levels, whatever they call them. The design rules below apply to any of them.

  1. Pick one stock unit per item, the smallest you transact. If you ever sell or count single bottles, the stock unit is the bottle, even if you buy only cases.
  2. Make every conversion item-specific. A global rule that “CS = 12” will be wrong for the first item packed in 10. Keep general conversions for fixed physical units such as pounds to ounces.
  3. Add supplier-specific purchase conversions where suppliers differ. Two suppliers can ship the same item in different case counts. Under the GS1 rule above, those are different trade items with different GTINs.
  4. Tie every conversion to a pack level. Each unit that represents physical packaging should have its own identifier, dimensions and weight. A unit with a conversion but no measurements will break putaway or rating later.
  5. Keep conversions whole where the item is not divisible. A fractional conversion is a sign of a modeling problem, unless the item genuinely sells by length or weight.
  6. Use the same unit codes everywhere. The ERP, the WMS and every EDI partner map need one agreed code per unit. Write down the translation for each partner’s codes.
  7. Decide rounding before go-live. If an order arrives for 30 eaches of an item stocked and picked in cases of 12, the system needs a rule: ship 24, ship 36, or break a case.

Variable-weight items

Some items cannot have a fixed weight: cheese wheels, meat cases, cut lengths of cable. GS1’s measurement standard notes that “the gross weight tolerances in these tables do not apply to variable weight trade items,” and points to attributes that allow a percentage of deviation from an average weight, or a minimum and maximum range, instead. In your systems, that means the item record holds a nominal or average weight for planning, and the WMS captures the actual weight of each unit at receipt or packing for invoicing and shipping.

Example (fictional). Item EX-20118 is a floor cleaner bought from two suppliers. Supplier A ships cases of 12; Supplier B ships cases of 10. The legacy ERP has one conversion, CS = 12, and the buyer adjusts Supplier B’s POs by hand. During WMS testing, an ASN from Supplier B for 40 cases is received as 480 eaches; the dock counts 400. The fix is a second purchase unit tied to Supplier B’s case GTIN, with its own conversion (1 case = 10 each), dimensions and weight. Receipts now convert correctly, and the stock unit, each, never changed.

Dimensions and weights: measure them the GS1 way

Measurements only compare if everyone measures the same way. The GS1 Package and Product Measurement Standard, release 3.2 ratified in March 2024, gives a repeatable method from the consumer unit to the pallet, and it is the reference most trading partners will recognize.

The rules that matter most for an item master:

  • Orientation first. For consumer units, the standard measures from a “default front.” While facing it, “Height: from the base to the top,” “Width: from the left to the right,” and “Depth: from the front to the back.” For cases and other non-consumer items, it starts from the “natural base,” described as “the natural underside of the packaged item pre-shipment.”
  • Maximum distance. The standard’s figure title says to “Always measure the maximum distance,” including projections, caps and lids.
  • Rounding up. “Millimetres are always rounded up to a whole millimetre,” and “Inches are always rounded up to the nearest 0.05 inch.” Its examples: 99.3 mm becomes 100 mm, and 2.932 inches becomes 2.95 inches.
  • Convert, then round. The conversion ratios are 1 inch = 25.4 mm and 1 mm = 0.03937 inches, and the best practice is “to first perform the conversion and then apply the rounding rules.”
  • One system for all three dimensions. The three linear dimensions are measured with the same measuring system and unit of measure.
  • Weight precision. The level of precision is left to the user and local law; if rounding is required, weights are rounded up. The conversions are 1.000 pound = 0.454 kilograms and 1.000 kilogram = 2.205 pounds.
  • Remeasure and sample. Measurements “SHOULD be retaken any time a change to the product or package occurs,” “Periodic audits SHOULD be conducted,” and “Multiple, random samples SHOULD be measured” to account for variation in manufacturing and packaging.

Two of those rules prevent the most common item master defects. Measuring the maximum distance stops a bottle’s cap or a box’s hang tab from being left out. Keeping all three dimensions in one unit stops the record where length is in inches and height in centimeters, which no automated check will catch unless you test for it.

Tolerances: how close is close enough

A tolerance is the allowed difference between the recorded value and a fresh measurement. Without one, every audit becomes an argument about whether a quarter inch matters.

The GS1 standard publishes tolerance tables for data accuracy. The figure shows the rows for common case types and the special rules for very small cases.

GS1 data accuracy tolerances for cases (selected rows)
Case typeDepth, width, heightGross weight
Corrugated case of cartons, cans, jars, rigid plastic or flexible packs±4.0% each±4.0%
Plastic overwrap of cans, jars or rigid plastic containers±4.0% each±4.0%
Soft paper products in flexible plastic or corrugated cases±10.0% each±9.0%
Flexible packaging in shrink-wrapped outer wrap±10.0% each±4.0%
Minimum when the percentage is 4%7 mm (0.25 in)0.1 kg (0.2 lb)
Very small cases: each dimension up to 160 mm (6.25 in)Fixed 7 mm (0.25 in)
Very small cases: gross weight up to 2.27 kg (5.0 lb)Fixed 0.1 kg (0.2 lb)
Source: GS1 Package and Product Measurement Standard, release 3.2 (March 2024), table 6-1 and section 6.2.1, checked November 2026. Consumer-unit tolerances are in a separate table of the standard and are not reproduced here.

The standard also states its limits. Its tolerances “apply only to fixed measure products,” and they “are not intended to replace parameters within the GTIN Management Standard, which identify when a new GTIN must be assigned.” Use the tolerance to decide whether a recorded value is accurate. Use the GTIN rules, below, to decide whether the product itself has changed.

In practice, a tolerance check is a simple comparison. GS1’s own worked example takes a small case with a depth of 190 mm and applies 4 percent to get a tolerance of plus or minus 7.6 mm, while its width and height, both under 160 mm, get the fixed 7 mm. If a fresh measurement differs from the record by more than that, the record fails.

Example (fictional). The case of EX-20118 is recorded at 12.4 x 9.3 x 9.9 inches and 11.6 pounds. An auditor measures three cases by the GS1 method and records the largest values: 12.5 x 9.4 x 10.6 inches and 11.9 pounds. At 4 percent, the tolerances are about 0.50 inch on depth, 0.37 inch on width, 0.40 inch on height and 0.46 pound on weight, all above the GS1 minimums. Depth, width and weight pass. Height is 0.7 inch over the record and fails. The change is about 7 percent, well under the 20 percent that would point to a new GTIN, so this is a record correction. The auditor notes the likely cause, a carton style with taller flaps, and the item’s height is updated with source “measured in-house” and the audit date.

When a change in measurements means a new GTIN

Sometimes a measurement is out of tolerance because the product changed, not because the record was wrong. GS1’s dimensional or gross weight change rule says: “A change to a physical dimension, on any axis, or gross weight over 20%, requires assignment of a new GTIN.” Its examples include a gross weight that rises 50 percent when packaging changes from plastic to glass, and a detergent box whose height grows 40 percent.

The same rule adds two points that belong in your item change process. “Frequent cumulative changes, without changing the GTIN, in avoidance of the 20% rule is an unacceptable practice,” and “Trading partners should be notified about all dimensional changes.” So an audit that finds a case 25 percent taller than its record is not a data fix; it is a question for the supplier about whether the item has a new GTIN.

How carriers and rating engines use these numbers

Freight is priced on size as well as weight, so item measurements flow straight into what you pay and what you quote. That is where measurement errors turn into money, usually weeks after go-live, on a freight invoice.

Parcel. Parcel carriers apply dimensional factors to package size, and the factor can differ by service and by customer. FedEx’s Ship API documentation, for example, notes a “different DIM factor of 4000 (for new FedEx customers)” on its intra-Europe Regional Economy services. Your rating engine holds the factor; the item master has to hold the true package dimensions and weight. The dimensional weight calculator shows the arithmetic and lets you test a package before you trust the system.

LTL. The National Motor Freight Traffic Association says the NMFC changes that “took effect July 19, 2025” bring “more detailed classification criteria for density, handling, and liability,” and that “your rating engines, TMS, and classification workflows should reflect these updates” (NMFTA). Density is weight divided by volume, so it depends directly on handling-unit dimensions and weight, pallet base included. The freight density calculator computes it from the same fields your item and pallet master should hold. Whether density sets the class depends on the item’s NMFC entry.

Example (fictional). A full pallet of EX-20118 measures 48 x 40 x 55.0 inches and weighs 741 pounds, as in the hierarchy figure. Its volume is 105,600 cubic inches, or 61.1 cubic feet, so its density is about 12.1 pounds per cubic foot. In the legacy system, the pallet height was recorded as 49.5 inches, the stacked cases without the 5.5-inch pallet base. That record gives 55.0 cubic feet and about 13.5 pounds per cubic foot, an 11 percent overstatement of density. On an item classed by density, that error can put the quote and the carrier’s inspection on different classes, and the difference comes back as a correction on the freight bill.

A second classic error is a unit mix-up. Suppose a supplier sends case dimensions in centimeters (31.5 x 23.7 x 25.2) and the import reads them as inches. The volume of that case becomes roughly 16 times larger, because one inch is 2.54 centimeters and 2.54 cubed is about 16.4. Every parcel quote for that case will be wildly high until someone looks.

Record where every value came from

Every measurement in the item master should carry its source, its date and its status, because the right fix depends on where a bad number came from. A supplier spreadsheet error is fixed with the supplier; a legacy guess is fixed with a tape measure.

Add these provenance fields to each measured attribute, or to each pack level if your system cannot hold them per attribute:

Field Example values Why it matters
Source Supplier data pool, supplier spreadsheet, measured in-house, derived, estimated Lets you audit by source and fix at the source
Source date The date the value was received or measured Old values are more likely to be stale
Method Tape and scale, dimensioning device, calculated from lower level A derived value inherits the errors of its inputs
Measured by Person or device ID Lets you trace a run of bad values to one setup
Samples Number of units measured GS1 recommends multiple random samples
Status Unverified, verified, failed, under review Drives the audit and blocks unverified items from rating if you choose
Next review A date or trigger, such as a supplier packaging change Turns “periodic audits” into a schedule

Derived values deserve a specific warning. If a case weight is calculated as 12 times the each weight, it ignores the carton, dividers and wrap, and it will be low. Mark derived values as derived, and replace them with measured values for any item that moves by parcel or LTL.

Which source wins when two disagree

It is common to hold two values for the same field: the supplier’s figure and a measurement taken at your dock. Decide in advance which one the systems use, so that the answer does not depend on who edited the record last.

A workable order of trust, from most to least: a recent in-house measurement by the GS1 method on multiple samples; a recent supplier value that has passed your audit for that supplier; any other supplier value; a derived value; an estimate. When a lower-trust value arrives for a field that already holds a higher-trust value, log it for review instead of overwriting. When the two differ by more than the tolerance, ask the supplier, because either your record or their product has changed.

Verifying item master data before go-live: a sampling audit

A pre-go-live audit has three layers: checks that run on every record, measurements on a sample, and tests of the live processes. The first layer is cheap and catches the gross errors; the second catches what arithmetic cannot; the third catches what nobody thought to check.

Layer 1: rule checks on the whole file

Run these against every active item before anyone picks up a tape measure. Each is a query, not a project.

  1. Completeness. Every pack level used by a live process has an identifier, a quantity, three dimensions, a weight and a unit for each.
  2. Units. All three dimensions of a level use the same unit, and the unit is one of the agreed codes.
  3. Ranges. No zero or negative values; no dimension or weight outside the plausible range for that level, such as an each heavier than its pallet.
  4. Cross-level weight. A case weighs at least its quantity times the each weight; a pallet weighs at least its case count times the case weight.
  5. Cross-level volume. A case’s volume is at least its quantity times the each volume; a pallet’s height is at least its layers times the case height.
  6. Conversions. The pallet quantity in eaches equals cases per layer times layers times eaches per case.
  7. Duplicates. No two items share a GTIN at the same level, and no item has two active rows for the same pack level and supplier.
  8. Process flags. Items that need lot, expiry or serial capture are flagged, and the flags match what the receiving team sees on the product.

Every failure goes on a list with an owner and a due date. Items that fail and do not move in the first weeks after go-live can wait; items on the fast-moving list cannot.

Layer 2: measure a stratified sample

A stratified sample groups items before sampling them, so that you learn which groups of data are unreliable rather than one blended error rate. Group items by data source (supplier data pool, supplier spreadsheet, measured in-house, legacy or unknown) and, within each, by velocity.

For each sampled item, pull units from stock, measure them by the GS1 method, and compare each value with the record using the tolerance rules. Record the result per field, not just per item, so you can see whether the weights are fine and the heights are not.

Sampled items out of tolerance, by data source Supplier data pool 4% (2 of 50) Supplier spreadsheet 14% (7 of 50) Measured in-house 2% (1 of 50) Legacy or unknown 38% (19 of 50) 0% 20% 40% Share of sampled items with at least one field out of tolerance
Fictional example data for illustration, using the tolerance rules from the GS1 Package and Product Measurement Standard.

Example (fictional). A distributor has 6,400 active items going into a new WMS. It samples 50 items from each of four data sources, 200 in all, and measures each at case and each level. The results are in the chart: 2 of 50 supplier data pool items fail, 7 of 50 supplier spreadsheet items, 1 of 50 in-house measurements and 19 of 50 legacy records. The team sets its decision rule before seeing the results: any source above 5 percent gets more work. In-house and data pool records pass. Spreadsheet items are sent back to suppliers with the failed fields listed. The 1,100 legacy items are ranked by velocity, the fastest 300 are measured in full before go-live, and the rest are flagged as unverified and measured at their first receipt in the new system.

The thresholds in that example are the distributor’s decisions, not a standard. Set yours before you sample, write them down, and get the operations and finance owners to agree. A rule chosen after the results are in tends to be chosen to pass.

Layer 3: test the processes with real items

Data that passes both checks can still fail in a process, because the process reads a field nobody audited. The last layer is to run each process from the table at the top of this note with real items and real units, in the test environment, and compare the result with what happens on the floor.

  • Receive a real supplier ASN and compare the converted quantities with a physical count.
  • Put away a full pallet and a mixed pallet, and check that the suggested locations fit.
  • Trigger replenishment on a pick face and check that it moves whole cases.
  • Pack a multi-line order and check the carton the system chooses against the carton a packer would choose.
  • Print each label type and scan it with the scanners your customers use.
  • Rate a parcel shipment and an LTL shipment, and compare with the carrier’s own quote.
  • Count a location in cases and confirm the posted quantity in eaches.

Write each of these as a test case with expected results; the UAT test plan template has the structure. A failed test is a defect against the data or the configuration, logged and retested like any other.

How bad item data shows up in the first weeks

If an error gets through, it usually announces itself as an operational problem rather than a data problem. Knowing the signatures saves days of debugging, because the floor reports a symptom and the fix is in a field.

Symptom on the floor or in the office Likely field First check
Every receipt of one supplier’s item is short by the same ratio Purchase conversion for that supplier Compare the case quantity on the ASN with a physical count
Putaway keeps suggesting locations that are too small for one item Case or pallet dimensions, or a unit mix-up Look for a dimension in the wrong unit or a pallet height without the base
The packing engine picks a carton that will not close Each or inner dimensions Measure the each by the GS1 method, caps and lids included
Parcel invoices for one item run well above quotes Package dimensions or weight Re-rate one shipment with measured dimensions
LTL freight bills come back with reweigh or reclass corrections Pallet dimensions, pallet weight or classification data Compare the bill of lading values with a measured pallet
A customer rejects labels or ASNs for one item GTIN at a pack level, or pack quantity Check the GTIN and quantity per level against the supplier’s current data
Counts post wildly high or low for one item Count unit and its conversion Confirm the unit the counter used and the conversion the system applied
Short-dated stock ships while newer stock waits Expiry control or shelf-life fields Confirm the flags and that expiry was captured at receipt

Treat a repeated symptom on one item as a data defect first and a system defect second. Fix the field through its owner, then check whether the same source produced other bad values, using the provenance fields described above.

Keeping item master data right after go-live

Clean data at go-live decays unless someone owns it and every change goes through one door. The work after go-live is a short list of controls, not another cleanup project.

  1. One owner per field. Decide which system and which team owns dimensions, weights, conversions and flags. The ownership model, and how changes flow between ERP and WMS, is the subject of ERP or WMS as system of record.
  2. A new-item gate. No item is released for receiving until its pack hierarchy, conversions, measurements and flags are complete and verified. A placeholder weight of 1 pound is the most common way bad data re-enters.
  3. Measure at first receipt. For items marked unverified, the receiving process captures case dimensions and weight at the first receipt and updates the record through the owner, not directly.
  4. Supplier change triggers. A supplier notice of packaging change starts a review: remeasure, check against the 20 percent GTIN rule, update conversions and partner maps.
  5. Scheduled audits. Re-run the rule checks monthly and a small measured sample quarterly, weighted toward fast movers and recent changes. GS1’s own text says periodic audits “SHOULD be conducted.”
  6. Integration monitoring. If item data flows from the ERP to the WMS by interface, watch for rejected item messages. A rejected update leaves the two systems disagreeing silently.

In our engagements, the item data rules, the new-item gate and the audit routine are written into the documentation and the training, so that the people who own the data after go-live are the people who learned how it is checked.

What to do next

Start from the processes going live and work backward to the fields. A practical order:

  1. List the processes in scope and the item fields each reads, using the table near the top of this note.
  2. Agree the pack levels, unit codes and stock unit for each item class, and write the conversion rules down.
  3. Add provenance fields for every measurement: source, date, method, samples and status.
  4. Run the layer 1 rule checks on the full file and assign every failure an owner.
  5. Sample by data source and velocity, measure by the GS1 method, and apply its tolerances with a pass rule agreed in advance.
  6. Test each process with real items and log failures as defects.
  7. Check freight math with the dimensional weight calculator and the freight density calculator for your top-shipped items.
  8. Put the gate and audits in place before go-live, then confirm the item master is on your go-live readiness checklist.

If the item data sits between an ERP, a new WMS and EDI partners, the interfaces that carry it need the same care as the data. Our ERP, WMS and EDI integration work covers data maps, error queues and monitoring for those interfaces, and you can contact us to scope an item master readiness review as part of a project.

Working on this problem?

If this field note describes a process you need to fix or build software around, tell us about it. We start with your process, your systems and your deadline.