You have a handle whose diameter could be anywhere between 1.1 and 1.7 inches, a screen that could sit at three angles, and a strap anchor that could be in four positions. The default plan is to print every combination and hand them out. Twenty-four models later you have spent three weeks and a few thousand dollars, and the answer is a dimension between two of the variants you built. The alternative — one prototype that adjusts through the whole range while a user holds it — usually costs less than four fixed models and produces a better answer, because the user can dial in what feels right instead of ranking things that all feel slightly wrong.
Adjustable prototypes are standard practice in seating, automotive interiors, and medical device human factors work, and badly underused everywhere else. The technique is simple: identify the two or three dimensions you are genuinely uncertain about, and build a model that varies only those.
When an adjustable model is the right call
| Build adjustable when | Build fixed variants when |
|---|---|
| Two or three continuous dimensions are open | You are choosing between distinct shapes, not dimensions |
| Users will evaluate feel, reach, or comfort | The test is appearance or finish quality |
| The range is wide and the optimum is unknown | You already have a narrow range and need a final check |
| You want data from many people | You need production-representative strength |
| Fixed variants would exceed four or five models | Two variants would settle it |
The pattern that pays best is the mid-stage decision: past concept sketches, before CAD is locked. It is also the right tool when a product must fit a range of people, because an adjustable model measures the distribution of preferred settings so you can pick a fixed dimension that serves most of it — how body measurement data becomes an actual dimension.
Mechanisms that make a model adjustable
- Slotted plates and thumbscrews. The workhorse. Laser-cut or machined slots with a knurled thumbscrew give continuous adjustment along one axis and cost almost nothing. Add an engraved or printed scale beside the slot so settings are readable.
- Telescoping tubes with a clamp. For length and reach. Aluminum tube in two sizes with a collar clamp is fast to build and stiff enough for most handheld tests.
- Extruded aluminum profile. T-slot framing with sliding nuts turns a bench fixture into a reconfigurable rig in an afternoon. Ideal for seating, workstations, and anything floor-standing.
- Detented pivots. A spring-loaded ball and a ring of dimples gives discrete, repeatable angle settings — better than continuous when users should compare three defined options.
- Shim stacks. Printed or cut spacers in known thicknesses change one gap or height in defined increments, and you can record exactly which stack the user chose.
- Interchangeable inserts. A common chassis with swappable grip shells or nose pieces captures shape variation that a slot cannot.
- Linear rails or slides. When a moving element must stay square while it adjusts, a small linear guide or bushing beats a slot, because a slotted joint racks under side load.
Most of these are assembled from catalog hardware, which is the philosophy behind building prototypes out of off-the-shelf components. A prototype full of knobs and rails does not look like a product, and that is fine — it is an instrument, not a sales model.
Designing the rig so the data is usable
An adjustable prototype is worthless if you cannot tell afterward what setting the user chose. Build measurement in:
- Mark every axis with a scale. Engraved lines, printed rulers, or numbered detents. Photograph the setting at the end of each session.
- Limit the range deliberately. Hard stops just outside the plausible range prevent users from wandering into settings you would never build.
- Change one thing at a time. If a user adjusts three axes simultaneously you learn what they liked, not why.
- Randomize the starting point. People anchor on wherever the model starts. Begin each participant at a different setting or you will measure your own assumption.
- Record the approach direction. Preferred settings differ depending on whether the user adjusted up or down into them. Doing both and averaging removes the bias.
- Keep mass honest. Adjustment hardware adds weight, and weight changes perceived comfort. Add ballast to a fixed reference model, or compare against a weight-matched dummy so the adjustable one is not penalized.
The last point is the method's biggest limitation: an adjustable model is heavier, bulkier, and less rigid than the product will be. It answers questions about dimension and position, not weight, balance, stiffness, or finish. Run the session as a proper usability test and tell participants what to ignore.
What it costs
A simple handheld adjustable model built from printed parts, aluminum tube, and thumbscrews typically runs $600–$2,500 including design time. A bench or floor-standing rig in T-slot extrusion with two or three axes is more like $2,500–$8,000. Compare that against fixed variants: at $200–$700 apiece for a finished printed and sanded model, six variants land in the same range and give you six data points instead of a continuous curve.
The savings compound across iterations. When testing reveals the interesting range is somewhere you did not build, a fixed-variant program starts over; an adjustable rig just moves the stops — the same argument behind most decisions about cutting prototype cost without cutting the information you get.
Where the approach breaks down
Skip the adjustable rig when the test involves real loads the adjustment hardware cannot carry, when the product's value depends on how it looks and feels as a finished object, or when the adjustability itself changes the interaction — a visible knob invites fiddling, and fiddling is not the behavior you want to observe. And if you need to know whether an assembly closes properly, an adjustable rig tells you nothing; that is a job for near-production parts and a tolerance stack-up analysis.
The best programs use both. An adjustable rig narrows a wide, uncertain range to a single dimension in one round. Two or three fixed models in that narrow range then confirm it and go to the people who approve the design. Deciding how many prototype units each round needs gets much easier once the wide search is done cheaply.
Projects House builds configurable test rigs — adjustable handles, seating bucks, reach fixtures — and runs the sessions that turn them into a dimension you can commit to in CAD. Tell us what you are still unsure about through the contact form.