In short: Warehouse wifi fails for reasons that are structural rather than accidental: steel racking blocks and reflects signal, stock levels change the radio environment week by week, and handheld scanners cling to distant access points instead of roaming cleanly. Manufacturers feel it as scanner dropouts, re-scans and inventory errors, and the fix depends on whether the building or the network is the real constraint.
Key Takeaways
- The racking is the radio problem — Steel uprights and palletised stock turn a warehouse into a maze of reflections and shadows, so a survey done at low stock is wrong by the next peak.
- Roaming is where scanners die — Handhelds hang on to a fading access point instead of switching, and every sticky handoff is a frozen screen and a re-scan mid-pick.
- Fix the diagnosis before the network — A compact site may need only a proper survey and more access points; a high-bay, multi-shed or yard-connected operation usually needs cellular.
In a nutshell
Why is warehouse wifi so unreliable?
Because a warehouse is close to the worst environment WiFi was designed for, and most warehouse networks were never engineered for it in the first place.
WiFi assumes short ranges, benign propagation and devices that stay put. A warehouse offers none of that. Steel racking runs in dense parallel rows, and every upright and every pallet of stock absorbs or reflects signal, so coverage is a patchwork of shadows that shifts as the racking fills and empties. Ceilings run to 12 metres or more, while the access points are usually mounted high on them and the scanners work at floor level, two racks deep in an aisle. Forklifts and VNA trucks are moving metal walls. And the devices that matter most, the handheld scanners and truck-mounted terminals, are in constant motion between all of it.
For a manufacturer this is not a side issue. Nearly every UK factory has a warehouse stapled to it: raw materials and components inbound, finished goods outbound, often a yard between them. The warehouse management system, whether it is SAP EWM, a Körber or Indigo installation, or a module of the site ERP, assumes every scan lands in real time. When the network drops the scan, the stock record and the physical stock part company, and everything downstream, from production planning to the customer's delivery promise, inherits the error.
What do wifi problems actually look like on the floor?
Rarely like an outage. Warehouse wifi fails retail, one interaction at a time, which is why it persists for years without becoming anyone's project.
The symptoms are familiar to anyone who has walked a floor: a Zebra or Honeywell scanner that freezes mid-pick and needs 30 seconds to reconnect; a putaway that has to be scanned twice because the first one never reached the WMS; a truck terminal that works in aisles one to six and drops in aisles seven to nine; a goods-in bay where labels get printed from a laptop because the wireless printer lost its connection again. Each incident costs seconds. Multiplied across every picker, every shift, the seconds become a measurable tax on throughput, and the workarounds are worse than the delays: operators batch their scans, or key entries in later from memory, and the real-time stock record quietly stops being real-time.
Then there are the errors that surface weeks later. A pallet putaway that never registered becomes a stock discrepancy at the next perpetual inventory count, which becomes a line-down event when production calls off a component the system says is in location 14-C and is not. Tracing that back to a dropped packet in March is effectively impossible, so the network never takes the blame, and the site concludes it has a "stock accuracy problem" instead of a connectivity one.
Why do scanners keep disconnecting when coverage looks fine?
Usually because of roaming, which is the part of warehouse wifi that heat maps do not show.
A survey measures signal strength at points in space. A scanner in use is a device in motion, and WiFi leaves the roaming decision to the device itself: the scanner decides when to abandon its current access point and associate with a nearer one. Handhelds are notoriously conservative about this. They cling to the access point they joined at the end of the aisle until the signal is nearly gone, then spend a second or more re-associating and re-authenticating, and the WMS session on top of that connection stalls or dies. The floor experiences this as random dropouts in areas the heat map shows as green, which is why so many warehouse wifi projects end with everyone frustrated: the survey was passed, the access points multiplied, and the scanners still hang at the aisle transitions.
There are palliatives, and a good integrator will apply them: tuning roaming thresholds on the device fleet, enabling fast-transition standards, thinning overlapping access points rather than adding more. They help, and on a modest site they may be enough. What they cannot change is the underlying design, in which every one of hundreds of client devices makes its own handoff decisions in a radio environment that changes with every stock movement.
Cellular handles this differently, and it is the single clearest technical argument for private 5G in a warehouse: the network, not the device, controls the handover between cells, and the handover is designed to be seamless because cellular was built for devices moving at speed. A scanner on a private 5G network crossing the site behaves like a phone on a motorway, which is to say the session simply stays up.
When is better wifi the right answer?
Genuinely often, and it is worth being straight about this because the answer shapes the budget.
If the site is a single compact shed, the racking layout is stable, and the pain is concentrated in a few known dead zones, then the right first step is a proper survey done at representative stock levels, access points repositioned into the aisles rather than spread across the ceiling grid, and the scanner fleet's roaming behaviour tuned. That is days of work, not months, and for many sites it clears the symptom list. WiFi also remains the natural home for guest access and office traffic whatever happens on the floor.
The case tips towards private cellular when one or more of the following holds: (i) the operation spans multiple sheds, a yard, or outdoor storage, where WiFi coverage between buildings is expensive and fragile; (ii) the building is high-bay or densely racked, so the access-point count and cabling cost climb steeply; (iii) the layout changes, because seasonal stock or a re-slotting programme invalidates any fixed survey; or (iv) the roadmap includes AMRs, automated trucks or vision systems, which need the deterministic performance and clean handovers that WiFi palliatives cannot promise. One or two small cells can cover a footprint, indoors and out, that would need scores of access points, on spectrum the site licenses from Ofcom for a few hundred pounds a year and shares with nobody.
The trade-off runs the other way too: cellular means SIM-capable devices (recent Zebra and Honeywell handhelds are, older fleets are not), and it carries more design work up front than adding another access point. A manufacturer with a five-year-old scanner estate and a stable single shed should probably sweat the WiFi first, and we tell them so.
What should a site do first?
Measure the problem where it actually occurs, which costs almost nothing. Pull the WMS transaction logs and look for the signature of a network problem: retries, duplicate scans, putaways keyed at a desk long after the timestamp of the physical movement, error rates that map to specific aisles or times of day. Walk the floor with the ops supervisors and ask where the scanners hang; the answer is always specific and always consistent. That evidence decides whether you have a tuning problem, a coverage problem or a structural one, and each has a different price tag.
We deploy and operate private 5G on exactly these sites, and the warehouse is frequently where a factory's network conversation starts, because it is where the pain is most visible and most easily counted. The wider prize is one network across the whole plot, production floor, warehouse and yard, carrying the scanners today and the AMRs and cameras next year. But it starts with the mundane question that brought you to this page: why do the scanners keep dropping in aisle nine? Walk the aisle, pull the logs, and the answer, and the right fix, will be in them.