Every Oil and Gas operator manages a spare-parts catalog that nobody fully trusts. The maintenance planner knows it. The procurement manager knows it. The reliability engineer who spent three hours searching for a pump seal before ordering a new one definitely knows it. But the problem does not appear on a dashboard, and it does not generate a service ticket. It accumulates silently, year after year, inside the item master.
The mechanics are consistent across upstream, midstream, and downstream operations. An acquisition brings a legacy SAP material master with 80,000 records using the acquired company's naming conventions. A platform expansion creates a new set of Maximo item records for the new storeroom. An emergency shutdown generates 200 procurement records that bypass the item-creation workflow entirely. None of these events are errors. They are normal operations. But their cumulative effect is an item master that carries between 8 and 18 percent duplicate or near-duplicate records.
On a 50,000-SKU catalog at an average inventory value of $850 per unit, that range represents $3.4M to $7.7M in physically redundant stock. On a large operator catalog of 200,000 SKUs, the exposure scales to $13.6M to $30.6M. These are working-capital figures derived from the catalog structure itself.
The turnaround context makes the problem acute. A major turnaround window lasting 14 to 90 days is one of the highest-cost operational events in an operator's calendar. Critical spares that appear unavailable trigger emergency procurement at premium rates. When the turnaround team cannot trust the catalog, they build shadow inventory lists, carry buffer stock that duplicates existing records, and buy from spot suppliers rather than preferred vendors. Industry estimates put turnaround procurement leakage at 3 to 7 percent of total turnaround spend for operators with unrationalized catalogs.
SAP material masters compound the challenge in a specific way. The MATNR-to-MAKTX relationship was never designed to prevent duplicate creation across plants. Plant engineers creating materials under their own organizational unit can produce records that are indistinguishable from existing entries to anyone who searches from a different plant code. Over a decade, a large operator can accumulate tens of thousands of plant-created duplicates that SAP native tools cannot reliably detect because the short-text descriptions are slightly different.
Maximo item catalog problems have a different character. Maximo stores item numbers, descriptions, and storeroom-level quantities without a global deduplication layer. Sites that manage their own Maximo instances create item records using local conventions. When those instances are merged, the duplicate density is typically higher than SAP environments — and the descriptions are more abbreviated and harder to normalize.
The diagnostic entry point is a CSV export. No ERP integration is required for the first step. The output must meet an industrial evidence standard: confidence-tiered findings, discriminator-controlled false-positive suppression, capital-at-risk quantification, and a governed review package that maintenance, procurement, and finance can act on together.
