Warehouse management software, explained for a company that has outgrown the spreadsheet

What a WMS does that an inventory spreadsheet cannot

A spreadsheet holds one number per product. That is genuinely enough for a long time, and anybody who tells you otherwise is selling something. It stops being enough at a specific and recognisable moment: when the number stops being the same for everybody who looks at it.

The moment usually arrives as reserved stock. You have 40 units on the shelf and 30 of them are already promised to orders that have not shipped yet. The spreadsheet says 40. Sales sells 40. Now somebody in the warehouse has to decide which customer gets told, and that conversation happens after the money has been taken.

A warehouse system splits that one number into three that are always true at once: on hand (physically here), reserved (spoken for), and available (what you can actually sell). Everything else a WMS does is downstream of being able to make that distinction.

The four signs you have outgrown the spreadsheet

These are the ones worth acting on. Each of them costs money in a way that is easy to miss because the cost arrives as a refund, a rush shipment or an hour of somebody's afternoon rather than as an invoice.

1. 01Two people can pick the same order. If there is nothing that claims an order when somebody starts walking it, eventually two people walk the same one — usually on your busiest day, because that is when there are two pickers.
2. 02You have oversold something you physically had. Not a stock error: the count was right, but part of it was already promised. This is the reserved-stock problem, and no amount of counting fixes it.
3. 03You cannot answer a recall question in an afternoon. If a supplier tells you lot 2411-B had a problem and you cannot list which orders it went out on, your only honest option is to contact every customer who bought that product.
4. 04The person who knows how the sheet works is a single point of failure. Every spreadsheet system eventually has one author and a set of conventions that live in their head. The system is fine until they take a holiday.

If none of these describes you, you do not need warehouse software yet. Buying it early is not harmless — it is a monthly cost and a process change in exchange for solving problems you do not have.

What to look for, in the order it matters

Feature lists all look the same at the comparison stage, because every vendor lists everything. These are the things that actually differ, roughly in the order that they will bite you.

• Does it run on the hardware your staff already hold? A picking screen that needs a rugged terminal is a purchase order and a lead time before you can start. One that runs on a phone camera starts on Monday.
• How does stock get reserved? Ask specifically when a reservation is made and when it is released. If the answer is vague, the oversell problem is not solved.
• Lots and expiry, if you sell anything that has them. FEFO — first expired, first out — has to be in the allocation logic, not a column somebody reads. If the system picks the wrong lot, a human noticing is not a control.
• What happens when a picker finds an empty bin? A system with no answer to this makes people abandon the job and fix the count later, which is how counts stop being trusted.
• Does it verify, or does it record? Recording what was packed is a log. Checking it against the order before the box is sealed is a control. Only one of those prevents a mis-ship.
• Can you get your data out? A CSV export and a documented API. This is the question that matters most on the day you leave, which is exactly when you have the least leverage to ask for it.

What it costs

Warehouse software splits into three price bands. Under about $50 a month you are buying an inventory tracker — a better spreadsheet with a stock count, and usually no picking, no scanning and no lot control. Between roughly $100 and $550 a month you are buying a real warehouse system aimed at small and mid-sized operations, which is the band Kinetel sits in. Above that, in the enterprise tier, pricing stops being published: you are quoted, the number is usually five or six figures a year, and it comes with an implementation project measured in months.

The trap in the middle band is the per-user charge. A system at $40 per user per month looks cheaper than one at $250 flat until you count the seasonal pickers, and warehouses hire seasonal pickers. Price it at your December headcount, not your February one.

We wrote the full breakdown — including the costs that are not on the pricing page, like data migration and the fortnight of reduced throughput while people learn — in a separate guide.

How long implementation actually takes

For a single warehouse with a few hundred SKUs and clean product data, a small system is running in days. The work is not the software: it is getting a product list with real SKUs, deciding what your bins are called, and counting what is on the shelves once, properly, so the opening numbers are true.

That last one is where projects actually slip. A go-live on top of a count nobody trusts produces a month of discrepancies that get blamed on the software, and the fix is expensive precisely because everybody has stopped believing the numbers. Count first. Go live on the count.

Enterprise implementations take months for a different reason — they are usually configuring the software to match an existing process rather than the other way round. That is the right call at a certain size and the wrong one below it.