The Parts Came Back Wrong. Now You Have to Change Supplier.
Moving a running part to a new shop is not a purchasing decision, it is a risk decision. You lose the fixtures, the programs and everything the old shop learned the hard way, and you only find out whether the new shop is better after the first batch lands. This page is what we ask for, what we do in the first month, and what we will tell you even when it is not what you want to hear.
Which of these went wrong matters more than the fact that it went wrong
A supplier that fails on dimensions needs a different answer from one that fails on dates, and one that fails on communication is the most expensive of the three. Before we quote a takeover we want to know which one you are dealing with, because two of them are not actually the supplier’s fault.
| What happened | What it usually means | What has to change in the takeover |
|---|---|---|
| First article passed, batch five failed | Nobody was measuring during the run. Tool wear moved the dimension and no one caught it until the customer did | Sampling through the run on the features that move, not just a first piece and a shipping note |
| Every piece slightly out on the same feature | A datum was picked up differently from the drawing, or the part is moving in the clamp | We re-read the datums against the drawing and design our own fixture — we never inherit a fixture concept that already failed |
| Dimensions fine, surface or finish rejected | Finishing treated as an afterthought, or a plating shop chosen on price alone | Finishing is specified and quoted as its own line, and we hold the supplier to it |
| Late every single time | Capacity, not capability — the shop took more work than it can run | An honest date at quotation, and a call from us the day we know a date is at risk |
| Something changed and nobody told you | Material substituted, an operation subcontracted, a revision missed | Certificates on request, no substitution without a written question first, revision confirmed on every purchase order |
| The drawing itself is not manufacturable | Sometimes the old shop was not wrong — they were quoting an impossible callout and hoping | We say so, in writing, before you place the order. Changing supplier will not fix a drawing problem |
Five things, and one of them matters more than the drawing
A rejected piece — two or three if you have them
This is the single most useful thing you can send us, and most customers do not think to. A part that failed tells us in an hour what a conversation cannot tell us in a week: where it moved, which way, whether it was consistent or random, and whether it was a machining problem or a drawing problem. Send them as they came back, not cleaned up.
The drawing, with the revision letter
Step or IGES plus a dimensioned PDF is ideal. If there have been revisions, we want to know which one the old shop was actually working to — a surprising number of takeovers turn out to be a revision that never reached the floor.
The inspection report that was accepted, if one exists
Seeing what was measured, and what was not, usually explains the failure on its own.
Real quantities and the delivery pattern
200 pieces once is a different part from 200 pieces every quarter. It changes the fixture we build, and it changes the price in your favour on the repeat.
What actually failed, in your words
Not the rejection code. What the part did, who found it, and how far down the line. If it reached your customer, say so — it changes how we plan the inspection.
What we do not want
We do not ask for the old shop’s programs, their fixture drawings or their process sheets, and we would not use them if you sent them. We write our own programs and build our own fixtures. Your competitor’s files would be treated the same way we treat yours.
What a takeover looks like from your desk
Drawing review before we quote
An engineer reads the drawing, not an algorithm. You get the price, the lead time, and the two or three callouts we think caused the trouble — including the ones we think are wrong on the drawing rather than wrong in the old shop.
Start with one part, not the family
If you have eight parts at a failing supplier, move the one that hurts most and keep the rest where they are. We would rather earn the other seven than be handed them on trust.
Our own fixture, our own program
Built here, proved here. Nothing about the old setup is carried across, including its assumptions.
First article measured, reported, and sent before the batch runs
Full dimensional report against the drawing datums, with the actual measured values — not a pass sticker. If a feature is marginal you hear about it at that point, not after 500 pieces.
Sampling through the run on the features that move
The dimensions that drift with tool wear get checked during the run, not at the end of it. That is the specific failure that sends most people looking for a new supplier.
One named engineer, for the life of the part
Bill, our engineering manager, answers the technical questions. Not a rotating sales inbox, and not a different person each revision.
Two jobs that show both halves of this
A first job in a material we had never cut
A driveline manufacturer sent us clutch pressure plates in compacted graphite iron — EN-GJV-300 / RuT300, machined from solid, twelve pieces, with a conical friction face held to a 0.14–0.23 mm taper and Ra 2.5–6.5 µm. It was our first job in that material and we said so before we quoted. Then we measured every single piece on the CMM, and had the friction-face roughness tested by an independent test house, rather than sampling, delivered in 15 days, and every piece passed the customer’s testing.
That is what a trial order should look like: the supplier names the risk before the order, and then over-inspects instead of hoping.
And a job we got wrong
Titanium knife scales for a customer of five years: to save cycle time we roughed and finished with the same cutter, and fine tool marks came through a surface the end user holds up to the light. We told them, we re-worked the batch by hand, we shipped five days late and paid for a discount ourselves, and we changed the process for every batch since.
We put that on the website because the supplier you are leaving probably did not tell you. How a shop behaves on the bad day is the only part of this that is worth checking.
Three things we will not do to win a takeover
Tell you the other shop was incompetent
We have not seen their machines or their drawing revision. Sometimes the part was quoted impossibly and everyone downstream paid for it. We will tell you what we see in the part and the print, and leave the verdict to you.
Quote a date we cannot hold to get you moved
The reason you are here is probably that someone did exactly that. A realistic date at quotation is worth more than a good one, and we would rather lose the enquiry than the second order.
Pretend a drawing problem is a supplier problem
If a callout cannot be held repeatably on any machine in any shop, changing supplier buys you one clean first article and the same failure at batch five. We will say that before you place the order, even though it is the fastest way to lose it.
Send one drawing and one rejected part
We will tell you what we think went wrong, what we would change, and what it costs — within 16 hours, at no charge. Start with the part that hurts most rather than the whole family. NDA on request before you send anything, and your files never leave the shop.