How Many Prototype Iterations Should You Expect?

Nobody gets it right first time, and the people who claim they do are not testing hard enough. The question worth asking is not "how do I avoid iterating?" but "how do I make each round cost almost nothing?" Get that right and the number of rounds stops mattering much.

The realistic number

There is no universal figure, but the shape of it is fairly consistent:

• A simple part — a bracket, a spacer, a mount, a cover — typically two to four rounds. Usually: one to find out you measured something wrong, one to fix the fit, one to confirm.
• A part with a mechanism — a hinge, a latch, a clip, a sliding fit — four to eight. Anything that has to flex and survive gets tuned rather than calculated.
• A part with a human hand or eye on it — a handle, a grip, an enclosure people judge — more, and less predictably, because "feels right" is not a dimension you can look up.
• An assembly — every added part multiplies the interfaces, and interfaces are where the rounds go.

If you are past a dozen rounds on a simple part, the problem is usually not the design. It is that each round changed several things at once, so nothing was ever actually learned.

Change one thing at a time

This is the whole discipline. If you adjust the clearance, thicken the wall and move the boss in one round, and the new one works, you know that some combination of three changes worked. You do not know which, you cannot undo the two that cost you material, and you have learned almost nothing transferable.

One variable per round is slower in theory and faster in practice. It also means a failure is informative rather than just annoying.

Print the variants together, not one after another

Here is the trick that makes the "one change at a time" rule cheap rather than expensive: you do not have to run those variants sequentially. Print the whole set in one go.

If you want to know which clearance works, do not print one part at 0.2 mm, wait, then print another at 0.3 mm. Print five, at 0.15 through to 0.55, on one bed, labelled. It costs a little more than one part and far less than five separate jobs, because setup and the machine's start-up overhead are shared and the bed is packed efficiently — the same effect described in small batch costs, and it is the reason quantity moves the per-part price. How pricing works sets out what actually drives the number.

Better still, do not print the whole part five times. Print the feature five times. A test coupon carrying just the hole, the boss or the snap — a few grams, minutes on the machine — answers the question exactly as well as a full-size part, and you keep the answer for every future design in that material.

Measure something, not "feels about right"

An iteration only counts if it produced a number or a decision you can write down. Before the print comes off the bed, know what you will do with it.

• Calipers, not fingers. "Slightly tight" is not data; 0.15 mm of interference is.
• Test the same way every round. If round two was tested by hand-pulling and round three on a scale, you have compared nothing. Where a fit or a position has to be checked repeatedly and consistently, a printed check fixture or go/no-go gauge pays for itself — that is what jigs and fixtures are for, and they apply to your own prototyping bench, not only to a factory.
• Test in the shipping material. A PLA test of a PETG part answers a question you did not ask.
• Write down what changed and why. Three months later, the reason you moved that wall 1 mm is gone, and someone will move it back.

When to stop

Stop when the next round has no question attached to it. Concretely:

• The last round changed something and nothing measurably improved.
• You are adjusting things nobody will notice — a 0.3 mm radius, a fillet on a hidden face.
• The remaining unknowns are ones a prototype cannot answer: will people buy it, does it survive a year outdoors, does assembly scale. Those need a batch in the field, not another print.
• The part does its job. Perfect is a tax that ships nothing.

At that point the design is frozen, and the money moves from prototyping to the first real batch. From prototype to production covers what happens next, and prototyping a product step by step sets out the arc that leads here.

Where another iteration is the wrong answer

Some problems do not get solved by printing the part again:

• The material is wrong. If a PLA clip keeps snapping, no amount of geometry tuning fixes brittleness. Change material — see how strong are 3D printed parts.
• The process is wrong. If you need a tolerance tighter than roughly ±0.1 mm or a genuine bearing fit, iterating on an FDM part is chasing something FDM does not reliably deliver. That is a machining job, and we will say so.
• You are guessing at a measurement. Another round of guessing costs the same as the last one. Measure the real thing, or send it to us.
• The design is finished and you are avoiding the commitment. That is not iteration, it is procrastination with a material cost.

Get an estimate · see the prototype printing service · upload a file