Everything has to land on the same day
Four parts of deciding, and none of them has a clock in it. This one does — because steel takes months, software takes weeks, and the data nobody planned for takes longer than either.
The argumentWhat you can see, and what you cannot
Racking is visible. You can point at it, walk around it, photograph it for the board. The system, the processes and the master data are not. There is no delivery lorry and no photograph, and for weeks there is nothing to show.
So they feel smaller than they are. They get less attention, they start later, and they are the reason a project does not deliver what it was built to deliver.
Not late — that is the smaller failure, and at least it is visible. The real one is a warehouse that goes live, works, and never reaches the throughput the whole case was built on. Nobody calls that a failure. The steel is up, the invoices are paid, orders go out of the door. It just costs more to run than it was supposed to, forever.
What it coversLead times, and the order they invert
Rough orders from projects we have run — our own planning figures, not published industry data.
| What | How long | What drives it |
|---|---|---|
| A vertical lift | 4 to 6 months | Build slot, delivery, then a floor that can carry it |
| Racking | Several months | Manufacture and installation; longer for anything non-standard |
| Conveyor, sorter | Months | Engineering before manufacture. The more technical, the longer. |
| The warehouse system | 6 to 12 weeks | Configuration, alignment, testing, training |
| Master data | Up to 4 weeks | Nobody's job, and it blocks the test |
| Scanners, printers, screens | Weeks | Until the network is not ready |
Read that against the order of the series. Part three decides the building and the equipment inside it — the four-to-six-month items. Part four decides the systems, which take a quarter as long.
So the thing you think about last is the thing that arrives soonest, and the thing you settled months ago is still in a factory. Order the steel when the software is specified and you have added four months for no reason: the lift does not care whether the WMS design is finished, only that somebody signed.
The part that saves monthsAlmost nothing actually waits
Most overruns are two things waiting on each other, and on inspection most of those waits are imagined.
Locations do not wait for racking. You have a drawing. Aisle, bay, level, position is a decision, not a measurement. What waits for the steel is the check that it stands where it was drawn — an afternoon, not a phase.
And testing does not wait for the warehouse. A table, a scanner, a label printer, a sheet of location stickers and ten articles. That is a working warehouse for testing purposes, and it will find almost every fault worth finding.
You find out that the batch decision sits in the wrong layer in week three, on a table, with ten articles — rather than in month five with the steel up and the people hired.
The full paper also covers the four weeks nobody plans — master data is not only the article file but packaging levels, location naming, suppliers, customers, batch rules — and why it is late in more projects than any other single cause. Plus the six tests that prove each layer is doing the job part four assigned it, and the instructions worth writing: not the full tray, but the short tray, the failed scan, the contested last unit.
The conclusionThroughput is not a property of any one of them
Equipment sets the ceiling. People do the work and change. Systems decide and direct. Processes make it repeatable.
A fast machine with a manual feed is a fast machine with a manual feed. A good system on wrong master data directs people to the wrong place with great precision. Neither is broken. Both are capped.
Part two set a target: a day's work inside a window. It is not met by the equipment, or by the software, or by hiring. It is met when the four line up — and the alignment is decided long before anybody is asked to hit a number.