How to measure supplier on-time delivery in a job shop
One late casting can quietly hold up four jobs. Most shops know which suppliers are a problem by reputation, but not by how much, on which parts, or which jobs it is costing them right now. The numbers to settle it are already in your purchase order history.
The basic formula
On-time delivery = deliveries received on or before the promised date ÷ all deliveries received
Four choices decide whether the number means anything:
- Which date counts as the promise. Use the date the supplier confirmed, not the date you asked for. Otherwise you are measuring your own lead times, not their reliability.
- Count lines, not orders. A purchase order with one late line out of ten is mostly on time, not late.
- Agree a grace period. Decide once whether a delivery a day late counts as late, and keep it the same for every supplier.
- Leave out what you can't measure. A line with no promised date or no receipt date is not on time. It is unknown, and counting it either way flatters or punishes the supplier.
On-time share is not enough
A supplier that is on time 90% of the time and three weeks late the rest can hurt you more than one that is always two days late. Track average days late alongside the on-time share.
Track rejects too: how much of what arrived was rejected or sent back, as a share of what arrived. Keep it apart from lateness. A supplier that is always on time with bad parts needs a different conversation from one that is late with good ones.
Grade suppliers so the answer is obvious
A ranked list of percentages invites arguments about the third decimal. Grades end them. These are the grades Shop Traveler uses:
| Grade | Means |
|---|---|
| Reliable | On time at least 90% of the time, and under a day late on average |
| Sometimes late | Between the two |
| Usually late | On time less than 70% of the time, or five days late or more on average |
| Not measured | Fewer than five measured deliveries: too little to say, never a clean record |
Go down to the part
Some suppliers are fine on most things and always late on one part. Grade each supplier's record per part as well, worst first. That is often the conversation to have with them: not "you're late", but "you're late on this casting, every time".
Example
A foundry is on time on 14 of 20 deliveries overall, 70%, so Sometimes late. Split by part, its brackets are on time 11 of 11, and its manifold castings 3 of 9, about nine days late on average. The brackets are fine. The castings are the problem.
Made-up figures, to show the method.
Plan on the record, not the promise
Once you know a supplier's usual lateness, add it to their promise when you plan. An order promised for the 10th from a supplier that averages six days late is due, for planning purposes, on the 16th. If the job needs it on the 12th, you know now, not on the 12th.
Tie late orders to the jobs they hold up
Purchasing's real question is not "who is late" but "whose lateness is costing us the most today". Link each open purchase order line to the job that needs it, and rank suppliers by the jobs, and the order value, waiting on their late lines.
How Shop Traveler does it
Shop Traveler reads your purchase orders every 15 minutes, read-only, and grades every supplier from its own delivery record, overall and per part, with its reject share beside the grade. The Suppliers page ranks them by how many jobs are waiting on their late lines, with the money at risk, and who to call. When a job waits on a supplier that usually runs late, the job is flagged early, showing both what the supplier promised and what to expect.