The hardware conversation usually happens too late. The software's picked, configuration is underway, and somebody in a status meeting asks what devices we're using. Two weeks before go-live, twenty tablets show up in a box.

That's when the warehouse finds out the back corner has no signal and the tablets can't read a label from more than eight inches away. A dropped one costs $600.

Hardware is the dull part of a WMS project. It's also the part your operators touch every minute of every shift, and it's where I see money wasted most predictably.

Warehouse Wi-Fi isn't office Wi-Fi

Your building probably has Wi-Fi people describe as fine. It's fine for a laptop that sits on a desk and can pause for two seconds without anyone noticing.

A handheld moves constantly. It hands off between access points at walking and forklift speed, and every scan is a round trip to the server. A two-second gap a laptop absorbs invisibly is a picker standing in an aisle tapping the screen.

Racking isn't drywall. Metal and dense product absorb and reflect signal. Coverage measured in an empty building isn't coverage in a full one, and January isn't peak.

Ceilings are high. An access point on a 30-foot ceiling puts most of its energy where nobody works. Lower mounting or directional antennas usually beat adding more APs.

Roaming matters more than speed. A device crossing between APs has to hand off cleanly. Consumer gear does this badly, and you see it as scans that hang at the end of an aisle.

Get a real site survey with the racks full. A predictive survey off a floor plan is a starting point. If your integrator will walk the building with the exact device model you plan to buy, that's the survey worth paying for. Make sure it covers the spots people already complain about, plus the freezer, the mezzanine and the dock doors with the doors open.

Ask about offline early

Ask what your WMS does when a device drops its connection mid-task. Some systems queue the transaction and sync when the signal comes back. Some fail the task and make the user start over. A few appear to accept the scan and quietly throw it away, which is the worst answer you can get.

If there are places you can't fix economically, like an outbuilding or a yard, the offline answer decides whether you can work there at all.

Rugged devices or phones

Phones are cheap and everyone knows how to use one. Their camera scanning is slower per scan, though, and it struggles with damaged labels and poor light.

Do the math per scan. If a picker scans 500 times a shift and the camera is a second and a half slower than a real imager, that's over ten minutes a day per picker, every day the system's in use.

Where I usually land:

  • Heavy scanning roles (picking, receiving, cycle counting): rugged devices with a real imager. They survive a drop onto concrete and the scan is instant.
  • Occasional scanning (supervisors, the office, a quiet second building): phones are fine, and the savings are real.
  • Forklifts: a vehicle-mounted terminal, or a rugged handheld in a holster. Not a tablet velcroed to the mast, which I've seen more than once.

Buy spares, about ten percent of the fleet and never fewer than two. A device out for repair with no spare means somebody works on paper, and paper doesn't reach the system the same day.

Check the battery spec against your real shift, with the scan engine and screen brightness running, and find out whether batteries swap hot. On a ten-hour shift it matters.

Labels and printers

Location labels are cheap until you buy them for a whole building. Get durable stock with an adhesive that suits your rack finish. Put a barcode and human-readable text on each, big enough to read from a forklift. A label someone has to walk up to and squint at is a tax on every put-away for as long as it's up.

For tall racks, add aisle signs and level markers so a driver can find a bin from the floor. If you're relabeling the building anyway, rename locations at the same time. It's the same ladder time.

On printers, check which label formats your WMS produces and whether your current printers can run them. Some systems expect a specific printer language, and your old units won't speak it. That's cheap to learn in testing and miserable to learn in go-live week.

Where the devices live

Decide where devices sit between shifts, who plugs them in, and what happens when one leaves in a jacket pocket.

Assign devices to people or stations and put a number on the back you can read across a room. The point isn't to punish anyone. Damage clusters, and the cluster usually points at something physical, like a rough ramp or a dock plate that bangs.

Put chargers where traffic already flows. If the charging rack is in the office and the pickers start at the far dock, the devices will end up charging in somebody's locker.

Order early

Rugged devices routinely run six to ten weeks out, longer with a specific radio or scan engine, and access points have their own lead times.

Order right after selection, not after configuration. You want devices in the building on the real network for testing and training, weeks before anyone depends on them. The first day an operator holds the device shouldn't also be the first day they're learning the software.