Material Selection for Product Design: Plastics vs. Metals vs. Composites

The right material for your product is the cheapest one that survives three tests: what the product has to do, how it will be manufactured, and how many you will make. Most first-time inventors run those tests in the wrong order, or skip them and design around whatever their 3D printer happens to use.

In 2026, material choice is still where a lot of hardware budgets quietly go wrong. A part designed as a 3D print and then "converted" to injection molding gets redesigned twice. A product specified in aluminum when polypropylene would do carries a cost penalty into every unit you ever sell. An outdoor gadget molded in unstabilized ABS chalks and cracks within a year. None of these are exotic failures. They are the default outcome of choosing a material by feel.

This guide gives you the decision process engineers use, a table to shortlist candidates, the manufacturing and cost logic behind each material family, and the eight material mistakes we see most in first-time projects. Every cost figure is a range, and the ranges match the ones used across the rest of this blog and in our product development glossary.

What material should my product be made of?

Pick by function first, manufacturing process second, and cost third, in that order. Run the three passes in sequence and the shortlist usually narrows to two or three candidates on its own.

Function comes first because a material that fails in use is worthless at any price. Process comes second because the process you can afford at your volume decides which materials are even available: injection molding opens the door to dozens of thermoplastics but demands tooling in the $5,000–50,000 range and a minimum order of 1,000+ units, while CNC machining handles metals and a handful of plastics with $1,000–15,000 in fixturing and viable runs of 100+ units. CNC stands for computer numerical control, and MOQ, the minimum order quantity, is the number that separates the two processes. Cost comes last because it only means something once the first two passes have told you which materials are in play.

There is a fourth filter that is easy to forget: the finished part has to look and feel like a product. A structurally perfect polypropylene housing that feels like a takeout container will hurt a premium product. Surface finish, color, paintability, and hand-feel belong in the function pass, not as an afterthought after the mold is cut.

The table below is the shortlist tool. Find your product in the "typical uses" column, check the process it needs, and read the watch-outs before you commit.

*Cost band combines resin or stock price, tooling implications, and unit cost at a typical first-run quantity. Low means commodity, High means specialty. Actual dollar ranges are in the cost section below.

Should my prototype be the same material as the final product?

Not for your first prototypes, and yes for your last one. Proof-of-concept and looks-like prototypes ($100–2,000 and $1,000–5,000 respectively; the four prototype types are covered separately) exist to answer questions about function and appearance, and PLA, PETG, or SLA resin answers those questions cheaply. The production-intent prototype ($5,000–30,000+) exists to prove the manufacturing approach, and that one has to be the real material made the real way, or it proves nothing.

This is the single biggest material mistake in first-time hardware, and it happens for an understandable reason: the prototype works, it is in your hand, and it looks finished. But the material in your hand is not the material in your product, and the geometry that printed cleanly may not mold at all.

The right way to use this table is to design your early prototypes knowing which production material they are standing in for, and to write that material on the drawing from day one. When you reach the production-intent prototype, build it in the real material using the real process, even if that means a soft aluminum tool or a machined-from-billet version of a part that will eventually be cast. The $5,000–30,000+ you spend there is what keeps the $5,000–50,000 production tool from being cut wrong.

One more consequence: if your prototypes were printed, your part geometry was designed for printing. Printers do not care about draft angles, uniform wall thickness, or undercuts. Molds care about all three. Expect a geometry pass, not just a material swap, when you move toward production. Our design-for-manufacturing guide walks through exactly what changes.

How do I choose a material by what my product has to do?

Write down every condition the product meets in real use and let each one strike materials off the list. Five conditions do most of the work: load, temperature, environment, contact, and feel. A material that fails any one of them is out, regardless of price.

Load and stiffness

Start with what the part carries and how far it is allowed to flex. Unfilled plastics flex a lot: a polypropylene bracket will bow visibly under a load an aluminum one would shrug off. Adding 30% glass fiber to nylon roughly doubles to triples its stiffness, which is why power-tool housings are glass-filled and not plain ABS. Aluminum has about one-third the stiffness of steel per unit volume but also one-third the weight, so an aluminum part designed to match a steel part's stiffness ends up thicker and lands around half the weight, not a third.

Two special cases are decided by material alone. Living hinges, the thin flexing strap on a flip-top cap, are polypropylene. Nothing else survives the fatigue at a reasonable price. Snap fits work in PP, ABS, and nylon; they fail early in brittle materials like standard SLA resin and acrylic, and they are a bad idea in glass-filled grades unless the geometry is generous.

Temperature

Temperature kills more prototypes than load does, because the numbers are lower than people expect. PLA softens around 130–140°F, and a closed car in summer reaches 130–170°F inside. ABS holds shape to roughly 175–195°F in continuous use, which makes it marginal in a dishwasher's heated dry cycle (up to about 180°F) and fine on a desk. Polypropylene is comfortable to about 210°F, which is why it fills kitchens. Polycarbonate runs to 240–260°F. Nylon spans 200–300°F depending on grade and glass content. Silicone is happy past 400°F. Aluminum and steel do not care about any temperature a consumer product will see.

Translate your use case into a number before you pick anything: dishwasher, 180°F; boiling water and steam, 212°F+; dashboard in July, 170°F; espresso machine internals, 250°F+. Then eliminate everything below it.

Environment: sun, water, chemicals

Sunlight destroys unstabilized plastics. ABS chalks and yellows within a season outdoors; if you like ABS, specify ASA, which is the UV-stable version of the same chemistry, or add a UV stabilizer package to PP or HDPE (a small resin surcharge, not a redesign). Polycarbonate yellows unless it is a UV-capped grade. Anodized aluminum and stainless steel are fine. Carbon steel rusts at the first scratch in the coating.

Water and humidity single out nylon, which absorbs 2–3% of its weight in moisture and swells and softens as it does. That is manageable in a hinge and a problem in a precision gear train. For anything that stays wet, stainless 304 handles kitchens and bathrooms; 316 is for salt water, chlorinated pools, and anything medical that gets aggressive cleaning.

Chemical exposure is the quiet killer of clear parts. Polycarbonate stress-cracks when it meets many solvents, some sunscreens, and a surprising number of household cleaners. Polypropylene and HDPE resist almost everything, which is why they are the default for bottles. If your product will be wiped down with alcohol or bleach, test the material with that exact product before tooling.

Contact: food, skin, and medical

"Food-grade" is a grade, not a polymer. Compliant grades exist for polypropylene, HDPE, PET, silicone, and specific nylons and polycarbonates; the resin supplier issues the compliance documentation for the exact grade, and the molder has to run that grade as virgin material rather than regrind. Specify all three on the drawing: the polymer, the grade, and "virgin only." Skin-contact and medical products add biocompatibility requirements that point toward silicone, medical-grade TPE, and stainless steel, and they add certification time to your schedule.

Feel and finish

Weight reads as quality. A zinc die-cast knob feels expensive because it is heavy; the same knob in ABS feels like a toy at a fraction of the cost. Decide which signal your product needs. Then check what the material allows: ABS paints and plates well, polypropylene resists paint and most adhesives, anodized aluminum delivers a premium surface straight from the process, and mold textures can hide the sink marks and flow lines that a glossy finish would expose. Soft-touch grips mean a TPE overmold and a second tool.

Which manufacturing process goes with which material?

The process you can afford at your volume decides the material family. Under 100 units you are machining, printing, casting in silicone molds, or bending sheet metal. Past 1,000 units you are injection molding, die casting, stamping, or extruding. The 100–1,000 band is where most first products live, and it is where the wrong call costs the most.

Under 100 units

CNC machining handles aluminum, steel, brass, and machinable plastics like acetal, HDPE, polycarbonate, and nylon, with fixturing in the $1,000–15,000 range and no minimum worth mentioning (see 3D printing vs. CNC machining for when each makes sense). Sheet metal covers steel, aluminum, and stainless with bend and weld operations and modest tooling. Industrial 3D printing in PA12 nylon produces functional parts that are close to molded nylon in behavior; SLS stands for selective laser sintering and MJF for Multi Jet Fusion, the two processes that make them. Urethane casting in silicone molds gives 10–100 plastic parts that look and roughly behave like ABS or PP, which makes it the standard bridge between a printed prototype and a molded product.

100 to 1,000 units

Sheet metal and CNC still work, and their unit costs barely move with volume. Soft aluminum tooling for injection molding starts to make sense toward the top of this band, especially when the part must be the production material for testing or certification. Machined metal is viable from 100+ units, which is the reason a metal first product can launch smaller than a plastic one.

1,000+ units

Injection molding is the default for thermoplastics, with a 1,000+ minimum order and tooling of $5,000–50,000. Die casting is its metal equivalent for zinc, aluminum, and magnesium, with tooling in the same range and the same volume logic. Stamping handles flat and formed steel parts. Extrusion produces constant-profile aluminum and plastic parts by the mile. Blow molding makes hollow HDPE and PET containers, thermoforming makes large thin parts, and rotational molding makes big hollow polyethylene shapes like coolers and tanks.

The rules each process imposes on your part

Injection molding wants 1–2 degrees of draft per side so the part releases, uniform walls between roughly 1.5 and 3.5 mm so it fills and cools evenly, ribs no thicker than 50–60% of the adjacent wall so they do not sink, generous radii, and no undercuts unless you are paying for side actions in the tool. Die casting wants the same plus tolerance for porosity. CNC wants internal radii that match a real cutter and tool access to every face. Sheet metal wants one gauge per part and bend radii at least equal to the material thickness. Read those rules before you finish the CAD, not after the first quote comes back.

One caution about instant-quote tools: they quote the process you selected. They will price 1,000 machined parts you should have molded without a word of objection. Process selection is an engineering decision, and the quote comes after it.

Plastic, metal, or composite: which is cheapest to make?

At 1,000+ units molded plastic is almost always cheapest, under 100 units machined metal or printed nylon often wins, and composites cost more in both material and process. The reason is that manufacturing cost has three parts, and the material price is the smallest of them for most consumer products.

The three costs

Material cost is the resin or metal stock itself. Commodity plastics (PP, HDPE, PET) sit at the bottom, ABS a step above, engineering plastics (PC, nylon, acetal) higher, and high-performance polymers many times higher again. Metals cost more per pound than plastics and weigh three to eight times as much per part. Composites add fiber, filler, and processing on top of the base resin.

Tooling cost is the mold, die, or fixture, and it is the number that surprises people. Injection molds and die-cast dies run $5,000–50,000 in 2026. CNC fixtures run $1,000–15,000. Sheet metal needs little beyond programming. What moves an injection mold from the bottom of that range to the top: the number of cavities, aluminum versus hardened steel, part size, undercuts that need side actions, tight tolerances, and cosmetic finish. Soft aluminum tooling at the low end is good for tens of thousands of shots; hardened steel at the high end runs hundreds of thousands to over a million.

Unit cost is what each part costs at your actual quantity, with the tooling spread across it. The arithmetic is where materials win or lose.

The crossover, with the reference numbers

Take a molded polypropylene housing with a $5,000–50,000 tool. At 1,000 units, tooling alone adds $5–50 to each part before any resin is bought. At 5,000 units it adds $1–10. At 25,000 units it adds $0.20–2. A CNC-machined aluminum version of the same housing has no such curve: the part costs about the same at unit 50 as at unit 5,000. So aluminum wins at 50 units and loses badly at 5,000, and the crossover for most consumer parts sits somewhere in the low hundreds to low thousands depending on part size and complexity.

A first production run for a small product costs $5,000–25,000; a medium run costs $25,000–150,000+; both take 8–16 weeks from purchase order to parts. Those figures are what factories charge for production, and they are separate from what you pay engineers. For the full cost picture, see how much it costs to develop a product and what a prototype costs.

What the material does to the tooling bill

The material you pick changes the tool you need. Glass-filled nylon is abrasive and pushes you to hardened steel rather than aluminum. High-temperature resins want heated tooling and hot-runner systems. Silicone needs a different molding process entirely. Multi-material parts (a hard housing with a soft grip) need two tools or a two-shot machine. Zinc die casting is heavy per part but its tooling and volume logic mirror injection molding closely enough that the choice between a molded and a die-cast knob is mostly about feel and weight, not tooling cost.

Composites deserve a plain statement: glass-filled nylon is a modest resin premium and a steel-tooling requirement, which is manageable. Carbon-fiber composite means hand layup labor or compression-molding tooling and materials that cost many times what plastics do. Reserve it for products where stiffness-to-weight is the product itself, like a drone frame or a racing component, and not for a housing that would look nice in carbon.

Where the tooling is cut matters too. Soft aluminum tooling from a domestic shop, ours in North Jersey included, is how you get real molded parts in the real material before committing to production steel, and it keeps the mold on domestic soil under your ownership. The ownership and lead-time trade-offs of domestic versus overseas tooling are covered in overseas vs. domestic manufacturing.

For scale: an inventor's total journey from idea through first production commonly lands between $25,000 and $225,000, and most of that goes to factories for tooling and parts, not to engineers. That figure is educational context, not a fee. Engineering fees for the design stage, where material selection actually happens, are a separate and much smaller number, covered below.

What are the most common material mistakes first-time inventors make?

Eight mistakes account for most of the material problems we see, and seven of them are decided before anyone calls a factory. Each one is avoidable with a sentence on a drawing or an hour of engineering time; each one costs a tool revision or a redesign if it gets through.

1. Designing for the prototype material. The part was drawn for a printer and it prints beautifully, so it feels done. Then the mold quote comes back with a list of undercuts, zero draft, and 6 mm walls that will sink and warp. Fix: decide the production material and process at the start of design, and let the prototype stand in for it deliberately.
2. Buying the premium material for a commodity job. Aluminum housings for a product that lives on a desk. Stainless where zinc-plated steel would last decades. Carbon fiber because it photographs well. Every one of these adds cost to every unit forever, and the customer usually cannot tell. Fix: let the function pass, not the mood board, pick the material.
3. Treating "plastic" as one material. ABS, polypropylene, polycarbonate, and nylon differ from each other more than aluminum differs from steel in the ways that matter: heat, chemicals, UV, stiffness, and what you can glue or paint. A design that works in ABS can fail in PP because the snap fit geometry is wrong for the stiffness, or succeed in PP and fail in ABS because the living hinge cracks. Fix: name the polymer and the grade, not the category.
4. Ignoring wall thickness and draft until the mold quote. Molded parts want uniform walls in the 1.5–3.5 mm band, 1–2 degrees of draft, and ribs at 50–60% of the wall. Parts drawn without these get quoted with side actions, extra cavities, or a redesign fee, and the tool lands at the top of the $5,000–50,000 range instead of the bottom. Fix: run a DFM pass on the CAD before you request a single quote. DFM stands for design for manufacturability, and the pass is a check of every part against the rules of its process.
5. Nylon without a moisture plan, or polycarbonate with the wrong cleaner. Nylon's 2–3% moisture absorption changes dimensions and stiffness; a gear train that meshed in the dry lab binds in a humid kitchen. Polycarbonate that will be wiped with alcohol or sunscreen-covered hands stress-cracks. Fix: match the material to the actual environment, and test the material against the actual chemicals before tooling.
6. Skipping UV stabilization on anything that goes outside. Unstabilized ABS chalks in a season. Unstabilized polypropylene embrittles. The fix costs a small resin surcharge (a UV package) or a material swap to ASA, and it has to be written on the drawing or the molder will run the cheapest natural resin available.
7. Assuming "food-safe" means the polymer, not the grade. Polypropylene is not food-safe; a specific polypropylene grade with the supplier's compliance documentation, molded from virgin material, is. Regrind, colorant, and process aids can each break compliance. Fix: specify polymer, grade, virgin-only, and the documentation you need, and keep the resin lot records.
8. Leaving the material off the drawing. A drawing that says "plastic" or nothing at all invites the factory to substitute whatever is cheapest that week, and the substitute is legal because you never specified. The material callout, the grade, the color, and the finish belong on your drawings, and those drawings, like every CAD file and specification produced in a properly structured engagement, belong to you. Fix: put it in writing, on the document you own, and make it part of the quote package.

How do engineers actually pick a material, and what does it cost?

Engineers pick materials by writing the requirements down, shortlisting two or three candidates, running a design-for-manufacturing pass on each, comparing cost at your real quantity, and then building a production-intent prototype in the winner. The whole process happens inside the design stage (Stage 2 of the five stages from idea to product), and it gets locked before any production tool is cut.

The sequence looks like this on a typical consumer product:

1. Requirements list. Loads, temperatures, environments, contact, regulatory needs, target unit cost, target quantity for the first run and for year two, and the look and feel the product has to deliver. This is a one-page document, and it is the most valuable page in the project.
2. Shortlist. Two or three materials that pass every requirement. Usually one commodity option, one engineering-grade option, and one alternative that trades a property for cost (glass-filled instead of metal, for example).
3. DFM pass. The CAD is checked against each candidate's process rules: draft, wall, ribs, radii, shrink, warp risk, gate locations. The candidate that forces the least redesign often wins here.
4. Cost at quantity. Tooling plus unit cost at the first-run quantity and at the year-two quantity, using real quotes rather than instant-quote estimates.
5. Production-intent prototype. The winner gets built in the real material by the real process, tested, and only then does the tool get ordered.

That is also how material selection fits into our approach at RMA: it is a design-stage deliverable inside our design services, it shows up on the drawings and in the specification package, and everything in that package is yours. Because we build prototypes in-house, the production-intent step happens at the same bench as the design, which removes the usual hand-off where a material assumption gets lost between the design firm and the shop.

On cost, in 2026 the industry range for product design work runs $3,000–25,000 depending on complexity, and material selection is a piece of that rather than a separate line item. RMA's fees start around $2,000, and basic products typically run $6,000–15,000 through design and prototyping to a manufacturing-ready package. Engineering time across the industry bills at $100–250 per hour. Those are engineering fees. The $25,000–225,000 total-journey figure quoted earlier includes production and is mostly paid to factories, and the two numbers should never be confused with each other.

Material decision checklist

Ten questions, in order. If you cannot answer one, that is the next thing to work on before spending money on tooling.

1. What load does the part carry, and how far may it flex?
2. What is the highest temperature it will ever see, including a car, a dishwasher, or a shipping container?
3. Will it see sunlight, water, or chemicals, and which ones exactly?
4. Does it touch food, skin, or a patient?
5. What does it need to feel like, and what finish does that require?
6. How many will you make in the first run, and how many in year two?
7. Which processes can make this shape at that volume?
8. What does the tooling cost for each candidate process, and what does each unit cost at both quantities?
9. Which candidate forces the least redesign of the current CAD?
10. Has the winner been built and tested in the real material by the real process?

Frequently asked questions

The short version

Choose by function, then process, then cost. Design your early prototypes as stand-ins for a production material you have already named. Build the last prototype in the real material by the real process. Put the polymer, grade, and finish on drawings you own. Do those four things and material selection stops being a risk and becomes the cheapest decision in the project.

If you have a product in design and want an engineer to look at the material question with you, book a Free 30-Minute Design Review. Bring the CAD, the prototype, or the sketch, whichever stage you are at.