Every WMS vendor has an ROI calculator on its website. Type in a few numbers and it tells you the software pays for itself in eight months. I don't use those, and I wouldn't let a client justify a purchase with one.

That doesn't mean the benefits are made up. There's published research on what warehouse systems and the discipline around them deliver, and when someone running Sage 100, Acumatica, NetSuite or Business Central asks me whether a WMS is worth it, these are the four numbers I reach for. Each has a source you can check.

1. A mispick costs about $22

The best public number on picking errors comes from a survey of 250 warehouse professionals commissioned by Intermec (now part of Honeywell). It put a single mispick at about $22, and the yearly cost of mispicks for an average distribution center at $389,000. The survey is more than a decade old, so treat those dollars as low. Freight and labor have both gone up since.

Two details matter more than the headline. Over half the respondents reported pick accuracy below 97 percent, and 19 percent didn't measure what mispicks cost them at all. In smaller operations, I'd bet that second group is bigger.

Your version is simple. Count last month's credits, reships and returns caused by a wrong item or wrong quantity, and multiply by $22 to $30 once you add labor and freight. That's your monthly bleed. I've walked into warehouses where that number alone justified a system, and others where it clearly didn't.

2. Best-in-class picking runs at 99.7 percent

The Warehousing Education and Research Council publishes an annual DC Measures benchmarking study from surveys of real operations. It's the closest thing this industry has to a neutral scoreboard. In recent editions, the top 20 percent hit order picking accuracy around 99.7 percent or better and inventory count accuracy of 99.5 percent or better. Typical operations sit well below that.

WERC doesn't say a WMS gets you there. Plenty of best-in-class warehouses got there with barcodes and disciplined cycle counting on top of their ERP's built-in warehousing. The benchmark gives you a target. If you're at 94 percent, the gap to 99.7 is real money, and a WMS is one of several ways to close it.

3. Manual processes land near 65 percent inventory accuracy

Auburn University's RFID Lab, working with GS1 US, tracked orders between brands and retailers in a study called Project Zipper. Orders shipped without item-level tracking had data discrepancies 69 percent of the time. With EPC tracking, order accuracy reached 99.9 percent. The same research community keeps finding that retailers on manual processes average around 65 percent inventory accuracy.

That number shocks people until they run their own wall-to-wall count. I've seen distributor counts come in anywhere from the low 80s to the high 90s, and the owners were surprised in both directions.

The takeaway isn't "buy RFID." For most smaller warehouses it's the wrong tool. The takeaway is that scanned, system-directed transactions beat memory and paper by more than anything else you can change.

4. Picking is about 55 percent of what the building costs to run

Academic work on warehouse operations, including the widely cited review by de Koster, Le-Duc and Roodbergen, puts order picking at roughly 55 percent of warehouse operating cost, and walking is the biggest share of a picker's day.

That's why directed picking is usually the biggest line in any believable WMS payback. Sequencing picks by location and batching orders goes straight at the largest cost in the building. You don't need a calculator for this one. Follow a picker for an hour with a stopwatch, and compare the time spent walking and searching with the time spent picking. That's your ceiling.

For context, Zebra's Warehousing Vision Study surveyed more than 1,400 warehouse decision-makers and found 85 percent have put mobile devices in workers' hands to capture moves as they happen. Your competitors aren't debating whether to scan anymore.

Rules for using these numbers

  • Your data first. The averages tell you the range. Your own error and count data tell you where you are.
  • Only count savings you can name. "Efficiency gains" isn't a line item. "Two fewer hours a day checking orders" is.
  • Cut every vendor claim in half. If the project only works at full vendor optimism, it doesn't work.
  • The studies measure discipline too. A WMS on top of sloppy receiving and no cycle counting produces precise, expensive records of bad data.

Do the math with your own numbers before you sit through a single demo. Whether a WMS pays for itself in your building is arithmetic, and the arithmetic is cheaper than the demo.