In machine building, assembly is an all-or-nothing moment. A fitter can only start once the kit is complete. If you have 199 of the 200 parts in on time, assembly still grinds to a halt. That one missing part, the shortage, determines whether the machine goes out the door on time. Purchasing in machine building is therefore not about ordering as much as possible on time, but about having exactly the right parts in house at exactly the right moment.
This sounds obvious, but in practice it is precisely where things often go wrong. Attention goes to order volume and to average delivery reliability, while the shop floor is waiting for that one casting, that one control component or that special bearing that has not arrived.
Assembly is only as on-time as its last part
Many purchasing dashboards measure delivery reliability per order line: what percentage of orders arrived on time. That number can look great, 95 percent or higher, while the build still stands still. The reason is simple. A machine is not built from individual lines, but from a complete bill of materials. One critical part that is late makes the whole kit late.
The relevant question is therefore not "how many parts were on time", but "was the complete kit complete before assembly had to start". That is a fundamentally different measure. Whoever measures at line level sees green. Whoever measures at kit level sees the reality on the shop floor. In supply chain terms: steer on on-time-in-full of the complete kit, not on the fill rate of individual lines.
Schedule backward from the assembly date, not from today
The order date of a part does not follow from "when did the demand arise", but from "when does it need to be there". That requires scheduling backward from the planned assembly start.
The reasoning runs from back to front:
- The customer has been given a delivery date.
- From that, the lead time for assembly, testing and shipping is subtracted.
- Then a buffer is subtracted, because the steps after the build also take time and something always goes wrong.
What remains is the date on which the kit must be complete. Subtract the supplier lead time for each part and you have the latest order date. For a standard bolt that is a formality. For a casting with a twelve-week lead time it means you have to order months before assembly, sometimes even before the design is fully finalized.
That built-in buffer is important to understand well. A part that arrives after the internal deadline is internally "late", but that does not automatically mean the customer is delivered late. The buffer is there precisely to absorb that delay. At the same time, the more of that buffer you use up, the less margin remains for the steps that follow. Keeping the internal deadline strict is not bureaucracy, it protects the delivery date to the customer.
Not every part carries equal weight
The biggest mistake is treating all parts equally. Your calendar is determined by a small group: the long-lead items on the critical path. These are the parts with the longest or most uncertain lead time, often also the parts that come from only one supplier.
In machine building these are typically:
- castings and forgings with long production times;
- welded structural parts and parts that still need surface treatment or finishing;
- special bearings, motors and drives;
- controls, electronics and semiconductors;
- customer-specific or certified components such as special valves or actuators.
For this category the rule is: identify them early, order them early, and monitor them separately. Standard C-items, bolts, nuts and seals are better covered with stock or kanban than with individual orders per machine. Your purchasing attention belongs where the risks are, not where the most lines are.
Where market signals give you a head start
Lead times and prices of critical parts are not fixed facts. They move with the market. If the lead time for semiconductors rises, the steel price increases, or a foundry becomes overloaded, your latest order date shifts forward without anything having changed in your own planning.
This is where the difference between reactive and proactive purchasing lies. Whoever only reacts when the order confirmation shows a later date is already too late. Whoever spots the market signals early can bring the order moment forward, activate a second source or temporarily increase stock on a critical part, before it hits the assembly planning. For long-lead items on the critical path, those few weeks of head start are exactly the difference between a complete kit and a machine standing still.
In practice
A workable approach can be summarized in a short checklist:
- Determine for each machine which parts are on the critical path and treat those with priority.
- Schedule order dates backward from the assembly start, not from the demand date.
- Keep the internal deadline strict, even when the customer buffer seems to absorb it.
- Measure kit completion and on-time-in-full, not average delivery reliability per line.
- Cover standard parts with stock, so your attention goes to the risky parts.
- Track market signals on your critical categories and bring order moments forward as soon as lead times rise.
In closing
Purchasing in machine building is not judged on how many parts arrived on time, but on whether the machine could be built on time. The two are not the same. Assembly only starts once the last part is there, so the slowest part on the critical path determines your success. Whoever recognizes that part early, schedules backward from the assembly date and stays ahead of the market makes sure the assembly department does not have to wait. And that, in the end, is what timely purchasing is about.