The Question Nobody Asks the Robot Vendor: Is Your Item Master Clean Enough to Automate?

Nobody asks the robot vendor whether your data can support the payback model, because the vendor's job is to sell throughput, not audit your catalog. That's a problem, because goods-to-person systems, AS/RS shuttles, and slotting algorithms don't execute your operation — they execute your item master, including every missing weight, wrong unit of measure, and broken case-pack hierarchy in it. Before anyone signs a check with seven zeros, someone should be running a dimensional-data fill-rate audit, and almost nobody is.

The capex gets a business case. The catalog doesn't.

Distribution Strategy Group has covered Associated Wholesale Grocers' $110 million commitment to automate its Gulf Coast distribution center — new equipment for ambient operations, aimed at order accuracy and throughput. It's one entry in a wave: the same outlet has tracked distributors expanding regional fulfillment footprints across 2026, largely to compress delivery windows and add resilience after several years of supply shocks. The projects get board approval, ribbon cuttings, and trade coverage.

What they don't get, in any of that coverage, is a line item for whether the underlying product data can carry the load. That's the gap. A distribution center automation business case models pick rates, storage density, and labor offset. It assumes the system will know how big and heavy each item is, in what unit, packed how many to a case. Most item masters don't reliably know that. They were built for a different job — ordering and invoicing — where a fuzzy or missing dimension record never stopped a truck from leaving.

What the robot actually reads

A picking robot or an AS/RS shuttle doesn't see the product. It sees the record. When the record is wrong, the machine doesn't pause to use judgment — it does exactly what a human veteran would have caught and corrected on instinct, except at machine speed and without the instinct. Zikoo Robotics has documented a version of this that shows up constantly in retrofits: a shuttle picks a pallet the database says weighs 800 kilograms, the pallet actually weighs 1,100, the load sensor faults, and the aisle stops. That's not a robotics failure. That's a data failure wearing a robotics costume.

The operations consultancy OpsDesign has made the same point more bluntly, arguing that data quality matters more than robotics selection: a robot sent to pick an item catalogued under three different SKU numbers, stored somewhere the system record doesn't match, or measured incorrectly in the item master will fail at that task no matter how well-engineered its arm is. Vendors sell precision. Precision applied to a wrong number produces a confidently wrong outcome, delivered faster than a person would have delivered it.

Distribution's own trade press has flagged the underlying readiness gap before it got fashionable to talk about robots. Distribution Strategy Group's 2022 piece on warehouse operations found two-thirds of distributors reporting satisfaction with their warehouse technology investment — while conceding the technology itself "does not come close to meeting" expectations, with only one in five very satisfied. Four years and a hardware upgrade cycle later, we'd bet that satisfaction gap traces back less to the equipment and more to what the equipment was asked to trust.

The slotting algorithm believes you

Cube-based slotting — grouping items by physical dimension to maximize density in a goods-to-person or AS/RS system — is one of the actual selling points of modern automation. It's also entirely downstream of dimensional accuracy. Feed a slotting engine a length, width, and height that's stale, estimated, or copied from a similar-but-not-identical SKU, and it will confidently assign that item to a bin it doesn't fit, or a zone that wastes the density gain the system was bought for. The algorithm isn't wrong. Its input is.

Same story with unit of measure. A robot told to pick "1 EA" against a record where the case pack is ambiguous — is this SKU sold each, or only in a 12-pack master carton that never got broken out correctly? — will pick the wrong quantity with total confidence. And packaging hierarchy — each, inner, case, pallet — is exactly the kind of structured, multi-level field that's tedious to populate and easy to leave partially filled, because a distributor's legacy PIM or ERP catalog was never audited against it. It's not that this data is technically hard to capture. It's that nobody built a workflow that forces it to exist before the automation RFP goes out.

The audit that should happen before the vendor tour

Here's the test we'd apply before any operator gets on a plane to see a robot demo: pull the item master fields the automation payback model actually depends on — weight, cube dimensions, unit of measure, and case/pallet pack hierarchy — and run a fill-rate audit against the full active SKU count. Not a sample. The whole catalog, because the automation system will eventually touch the whole catalog, including the long tail nobody looked at during due diligence.

If that audit comes back with dimensional fill rates in the 60s or 70s percent — typical for a distributor catalog that's grown by acquisition, the pattern we see across the six operating archetypes in our own Top Distributors 2026 index — the payback model isn't wrong on its math. It's wrong on its assumption. It assumed data accuracy the item master doesn't have, and the gap between assumed and actual accuracy is exactly where automation ROI quietly leaks: faults, re-slots, manual overrides, and a go-live that runs slower than the manual process it replaced, for months, while the data gets fixed live under production pressure instead of ahead of it.

This is measurable before a single robot ships. Our Digital Readiness Index methodology scores distributors on structured-data completeness for exactly this reason — dimensional and packaging fields are signals of operational readiness, not just ecommerce polish.

The fix isn't a new system of record. Your PIM or ERP stores the data; it doesn't have the throughput to close a 30-point fill-rate gap on its own. That's the work Anglera does — starting from a flat export if that's all you have, live in weeks rather than quarters, closing dimensional and packaging gaps at scale before the automation vendor arrives, not after the shuttle faults on a pallet that's 300 kilograms heavier than the record said.