It happens in a predictable way. An engineer or a technical founder builds the thing that works first, because that is the risky part. The mechanism functions, the board boots, the firmware runs, and there is a working unit sitting on a bench. Then somebody shows it to a potential customer or an investor and the reaction is polite. The product works and looks like a science fair project, and now the question is whether industrial design can still fix it without throwing away a year of engineering.
The answer is usually yes, partially, and the size of "partially" depends entirely on what has already been frozen. This is triage: what a designer can still change for almost nothing, what costs a re-spin, and how to sequence design and engineering next time.
First, find out what is actually frozen
Before any design work starts, someone has to write down the real constraints. Not the assumed ones. In most projects that arrive in this state, about half of what the team believes is locked is not locked at all — it was a default nobody revisited.
Make an explicit list under three headings:
- Hard-frozen. Tooling already cut. Certified board layout that passed emissions testing. A component with a 40-week lead time already on order. A regulatory submission already filed against a specific configuration.
- Expensive to change. PCB outline and connector positions, mechanism kinematics, motor and gearbox selection, battery size, anything that would restart a qualification test.
- Actually still open. Surface geometry outside the electronics envelope, split lines, textures, colors, graphics, button caps and light pipes, packaging, and almost everything a user touches.
That third list is bigger than teams expect, and it is where a designer earns their money on a salvage job.
What can still change cheaply
If the internals are fixed but the housing has not been tooled, you are in good shape. The industrial designer is designing around a known internal volume rather than inventing one, which is constraining but not fatal — this is roughly how electronics enclosure design works even in a clean project.
Cheap changes at this stage typically include:
- Outer surfacing. The skin can be completely redrawn as long as it clears the internal envelope with the wall thicknesses and boss locations engineering needs.
- Split-line strategy. Moving a seam so it lands on a design feature instead of across a face is often free before the mold is cut, and it is one of the highest-value visual changes available.
- Texture and finish. Mold texture is applied late in tool build and costs a few hundred to a few thousand dollars. Choosing well matters more than most founders think; see mold texture selection.
- Color, graphics, and branding. Pad printing, in-mold labeling, or a molded logo can be added late. Our guide to putting a logo on a product covers what each method costs.
- Button caps, light pipes, and bezels. These are usually separate small parts. Redesigning them is cheap and changes the perceived quality of the product dramatically, because they are what the hand and eye land on.
- Packaging and the unboxing sequence. Completely independent of the electronics, and often the single fastest improvement to how the product is received.
What triggers a re-spin, and what it costs
Some design requests look small and are not. These are the expensive ones, roughly in order of pain.
| Change | What it really touches | Typical cost impact |
|---|---|---|
| Move a connector or a USB port | PCB layout, possibly EMC retest | $3,000–$15,000 plus 4–8 weeks |
| Change board outline or thickness | Full layout re-spin, new assembly setup | $8,000–$40,000 plus 8–14 weeks |
| Move a status LED or a button | Layout change, or a light pipe and flex cable workaround | Often solvable mechanically for under $2,000 |
| Change an antenna location | RF tuning, recertification of the radio | $10,000–$30,000, schedule risk |
| Modify a mold that is already cut | Steel welding and recutting, or a new insert | $1,500–$20,000 per change; adding material is far cheaper than removing it |
| Change overall proportions after tooling | New tool | Full tooling cost again |
The mold rule deserves emphasis because it is counterintuitive. In a cut steel tool, making a plastic part thinner or smaller generally means adding steel, which is cheap. Making the part thicker or bigger means removing steel that is already gone, which means welding and remachining or a new insert. A designer who understands this can often get most of the visual improvement they want inside a tool that already exists. The geometry that decides this is covered in draft angles and parting lines.
The parting line problem
The most common visual complaint about engineer-first products is not the shape. It is the seams. Parting lines placed for molding convenience run across the middle of a face, ejector pin marks land on a visible surface, and the gate leaves a witness mark somewhere a customer's thumb rests.
If the tool is not cut, this is a free fix and it should be the first thing on the list. If it is cut, you have three fallbacks: hide the seam under an applied part such as a rubber band or an overmolded grip, break the surface deliberately with a groove or step so the seam reads as intentional, or accept it and put a texture on the face that makes it less visible. All three are used constantly.
What to do when it is genuinely too late
Sometimes tooling is cut, units are shipping, and the product still looks wrong. That is not a disaster, it is a version-two brief. Ship what you have to the customers who care about function, collect real feedback, and put the design work into the next revision where it can shape the architecture instead of decorating it. Our article on when to build version 2 covers how to decide the timing. In the meantime, packaging, graphics, and accessories are the levers you still control.
If you do change the design of a shipping product, run it through a controlled process rather than a hallway decision. The mechanics are in the engineering change order process, and skipping it is how a company ends up with three incompatible versions of the same part in the warehouse.
Sequencing it properly next time
The reason this happens is almost never laziness. It is that function felt like the only real risk. The fix is not "do design first" — that produces concepts nobody can build. Run the two in parallel with defined checkpoints.
- Design and engineering start together on the same brief, with the designer sketching while the engineer runs feasibility on the mechanism or the radio.
- Freeze the internal architecture and the external form at the same review. This is the single most important checkpoint. The board outline, the mechanism envelope, and the outer surface get approved as one package.
- Physical mock-ups before any tooling. Print the housing, put the real board and battery inside, hold it. Errors that cost $50 to find here cost $50,000 after tooling — the case for this is made in what to check in the first mold trial, by which point your options have already narrowed.
- Designer stays through DFM. Not a handoff. The designer should be in the room when the molder asks for more draft, because that is where the form quietly degrades.
The working relationship that makes this possible is described in how engineers and industrial designers work together, and it is worth reading before you staff the next project.
Projects House regularly picks up products that work and need to look like they belong on a shelf, and we will tell you honestly which changes fit inside your existing tooling and which do not. Send us photos and CAD of what you have through the contact form.