Define items and operational records

Knowing where a case moved is not enough if the “item” in purchasing differs from the item in sales. An item is an operational definition of something that can be counted, ordered and stored consistently. Two packs may have different sizes, versions or handling needs despite sharing a commercial name. The team chooses enough detail to prevent noninterchangeable products from being combined in a decision. Location matters in the same way: “the warehouse” is too broad when one unit is quarantined and another is on a saleable shelf. To build a usable record, begin with four questions: what is counted, in what unit, where is it, and in what status?

Prepare a small data dictionary before importing a table: item ID and description, base unit, permitted pack units and each conversion factor, location ID, stock status and record time. Separate physical quantity from promiseable quantity. Physically present stock may include a return awaiting inspection or units reserved for another order. Stock status must prevent an invalid sum. GS1 explains how product and location identifiers connect physical and information flows. Use of actual global identification keys follows the issuing system's rules. Our training IDs are explicitly fictional; we do not claim they are valid GTINs.

A practical reconciliation connects three records: the purchase order with its unit, the physical and accepted receiving quantities, and the inventory movement that changes the balance. Convert each quantity to the base unit, then compare. Check identity before arithmetic: is it the same SKU, or only a similar name? Check the pack specification effective when the order was placed, rather than today's factor if packaging changed. Then separate count difference from status difference. Three hundred units may arrive physically while only 288 are accepted because one case is damaged. The right event record is “300 received, 288 accepted, 12 isolated,” not “12 short-shipped” until inspection resolves the cause. This separation preserves the decision trail and keeps isolated goods out of promises.

Some data errors cannot be repaired by conversion. Systems might record L-20 and L20 as two identifiers for one item, dividing the balance and making one appear out of stock. Before merging them, compare product number, pack, supplier, specification and transaction history. The identifier difference might hide a real composition or shelf-life change. An integration message can also post the same receipt twice. Compare movement ID, source and time rather than deleting one row because its quantity matches. Corrections require a data owner, authority and audit trail. The specialist prepares an exception list and demonstrates its effect on an operational decision.

Finish with a concise exception record: item ID, conflicting records, original and base units, conversion used, calculated variance, evidence needed, effect on promise or purchase, and correction owner. Test the dictionary with a small end-to-end movement: order one case, receive 12 eaches, sell two, and explain a ten-each balance if there are no reservations or rejections. A correct sum alone is insufficient. Quantities must sit under the correct item, location and status. The test turns the dictionary into an operating tool rather than a list of terms.