The frustrating thing about a sample is that it consumes a large share of a project's elapsed time and delivers one object. A buyer who has approved a drawing reasonably expects the sample to be quick, because the part looks simple and the batch that follows will run on the same tooling. The batch usually is quicker. That is not because the shop works harder on the sample. It is because the sample is where every unresolved question in the project has to be answered.
A batch has one answer per question, applied repeatedly. A sample has to produce those answers first, and the work of producing them does not look like manufacturing. That is why it is so consistently underestimated in a schedule.
Some of the elapsed time on a sample is cutting, forming, welding and finishing. Most of it is not. It is reading the drawing and finding the places where two views disagree, or where a dimension is given to a feature that does not exist yet. It is deciding which of two possible readings the customer meant, and asking. It is waiting for that answer.
That waiting is the part of a sample nobody schedules. A shop can compress machine time by working later. It cannot compress a question that has been sent and not answered, and it cannot usefully start cutting metal against a geometry that may still change.
The two ways a custom job starts look different but share a bottleneck. If the customer supplies a drawing that carries dimensions, material and special requirements, the mould can be made from it directly. If the customer sends a sample instead, the sample goes to an engineer, who measures the dimensions and produces a drawing, and the mould is made from that drawing.
The drawing route looks shorter, and in a sense it is, but only because the measuring work has already been done by someone else. The sample route makes that work visible. Neither route skips the step where the shape has to become unambiguous.
Once the geometry is settled, making the tooling is the longest single operation in the sample stage and the one least able to be hurried. A mould has to be cut, fitted, polished wherever the surface matters, and then tried. Trying it produces a part, and measuring that part produces either a confirmation or a correction.
This is also where the choice of tool matters. A sample mould is smaller and cheaper and makes samples only. A big mould makes both samples and mass production. The distinction exists because the two jobs have different economics: a sample mould lets a project prove a shape without committing to the tool that will run the batch, which is the right call while the shape is still moving.
The consequence is that a project can pass through a sample mould and still need a second tooling stage before volume production. That is a sensible trade, but it is one a buyer should make deliberately rather than discover halfway through.
A mould is only half of what has to be right. The part also has to be held for every operation that follows it: trimming, machining, welding, finishing. Each of those needs its own fixture or its own position, and each one is a place where a part can move.
On a first sample those fixtures are being used for the first time. A fixture that holds a part slightly off plane will produce a part that measures correctly in one place and pulls somewhere else, and finding that out takes a fitting check rather than a drawing check. That check is quick to describe and slow to do, because it is the one where the shop discovers its own assumptions.
The process is worth following as a sequence of gates: the mould is tested, samples are made, and an engineer measures sizes and tests hardness. The samples then go to the customer, and mass production starts after confirmation. Each of those gates costs time, and each one can send the job back a step.
Measuring a sample is not a single act. It is the dimensions on the drawing plus the dimensions that are not on the drawing but matter: flatness across a flange, the way two mating faces sit together, the wall thickness left after forming. Hardness testing adds another operation and may be destructive, which means it consumes a sample rather than returning it.
Two customer behaviours move a sample date more than anything the shop does. The first is changing the requirement after the sample exists. The second is approving slowly, or approving in conversation and raising a point later.
Both are normal and neither is unreasonable, but both reopen a gate. A change that touches geometry restarts tooling. A change that touches only finish or colour restarts finishing. The schedule does not weigh the size of the change. It only knows that something already closed has been opened again.
A sample stage can be shortened, and shops do it when a customer insists. The usual method is to skip the formal measurement gate and move to production on the basis that the sample looked right. It works often enough to stay tempting.
When it fails, it fails across the whole run, because the unresolved question has been reproduced on every part in the batch. The time saved at the sample is then spent on sorting, rework and remade tooling, and it is spent at the worst possible moment, which is after the customer has already promised a date to somebody else.
The useful way to look at a sample date is as a statement about how many decisions remain open. If the drawing is closed, the material is chosen, the finish is specified and the approval route is agreed, then the sample stage is mostly work, and work can be planned. If any of those is still under discussion, the date is mostly intent.
This is why two identical parts can have very different sample stages. The difference is rarely the shop. It is how much of the definition arrived with the enquiry.
It also helps to know which of those decisions sit with the shop and which sit with you. The list is usually shorter than buyers expect, and closing it is the only thing that reliably shortens a sample stage.
Each of those is a gate that closes once. A project that closes them early has a schedule. One that leaves them open has a conversation.
A shop that tells you which decisions it is waiting on is more useful than one that gives you a date without conditions. If you want to see how that is handled here, the customization page lists the inputs we ask for and the order in which they are needed. The plant page describes the workshop where tooling is built, fabricated titanium components shows the geometry that most often needs a fresh mould, and other parts covers the smaller items where the sample stage is short because the shape is simple. Buyers running a programme of parts can read the dealership route, and Tigindustry Swap is where an existing shape gets reused instead of retooled.
