Clean the Item Master

Duplicate SKUs and inconsistent units quietly break order automation, reorder logic, and your inventory numbers. Cleaning the item master is the cheapest high-return fix most distributors skip.

The problem, in your words

The same part lives under three codes. Units don't agree — eaches here, cases there, no conversion in between. Some items are missing a cost, a lead time, or a description a human could actually read.

It mostly works. Your people know the workarounds — which code is the "real" one, which vendor item maps to which. But automation doesn't know the workarounds. And neither does your inventory report.

Why it persists

The item master grew over decades — through ERP migrations, an acquisition or two, and whoever set up new parts that week. Cleaning it has no deadline and no owner, so it never rises to the top of anyone's list. The cost is hidden in plain sight: split inventory, broken reorder points, orders the system can't match, reports nobody fully trusts. None of it shows up as a line item called "messy data."

The teach — why it's load-bearing

Order automation works by matching an inbound line to an item record. Duplicates make that match ambiguous, so those orders fall to a human — which is exactly why your touchless rate caps out below where it should.

Reorder logic splits demand across duplicate codes, so you stock out of an item you technically "have," sitting under a different number. And GMROI and turns are computed on records that double-count, so the metrics you steer by are quietly wrong.

The item master isn't admin. It's the floor everything else stands on. You can't automate, forecast, or trust a number built on a record that's three records pretending to be one.

The play

1. Profile it. Duplicate rate. Null rate on key fields (cost, UOM, lead time, description). UOM consistency by category. One scan, real numbers — not a vibe.
2. Rank by impact. Clean the fast-moving, high-revenue, most-ordered items first. You don't need a perfect master. You need your top items clean.
3. De-duplicate with a rule, then a human. Auto-cluster likely duplicates; let a person confirm the merges. Keep an audit trail so nothing disappears silently.
4. Fill the load-bearing fields. Cost, UOM and conversions, lead time — the fields automation and reorder logic actually read.
5. Hold the line. A short intake standard for new items, so the mess doesn't grow back the week after you finish.

How to measure it

• Duplicate rate and null rate on key fields — before and after.
• Share of your top-N items that are clean.
• Then watch the downstream numbers: touchless rate up, order exceptions down, reorder points you can finally trust.

The number it moves

Gartner finds 70%+ of ERP initiatives fall short of their business case, frequently because of data. You can't see this fix directly — there's no "item-master savings" line. You see it in everything downstream: more orders matched touchlessly, fewer exceptions, inventory math you can act on.

This is the paid first step that de-risks every automation after it. Skip it, and you're automating a mess at scale. Do it, and the rest of the ladder gets cheaper and more accurate.

[PLACEHOLDER: MarginArc client item-master scan results — to be added]

Nobody wants to clean the item master. Everybody pays for not doing it.

Every number here is either yours (the calculator, the audit) or an attributed benchmark. Talk to us about your number →