NRE: design, fixtures and first build, and what you divide them by
At low volumes the one-time costs do not get thin. How to identify them, how to choose the quantity to spread them over, and why ignoring them makes a quote look competitive and lose money.
In high-volume manufacturing, one-time costs get divided by a huge number and effectively vanish. At the volumes this trade runs at, the same costs divided by a small number do not vanish at all — they can exceed the material.
This is where machine builders lose money, and the error runs in one direction only: **the real cost is always higher than the quoted one.**
Which costs are non-recurring
One test settles it: **if you built a second unit, would you have to do this again?** No means non-recurring.
| Cost | Repeat on unit two? | Verdict |
|---|---|---|
| Design and drawings | No | Non-recurring |
| Special fixtures and jigs | No | Non-recurring |
| CNC programs | No | Non-recurring |
| First build, debugging, getting it to work | No | Non-recurring |
| Material and bought-out parts | Yes | Per unit |
| Assembly labor | Yes | Per unit |
| Acceptance test on each unit before shipping | Yes | Per unit |
The hard question: divide by how many
By the units in this order, or by the units you hope to sell? Both are defensible and they give very different answers.
- This order only — conservative, higher unit price, and you can never fail to recover the cost. Right when the machine is specific to one customer.
- Expected future volume — lower unit price, more competitive, but you are betting on orders that do not exist yet. If the second order never comes, the unrecovered part is a real loss.
When it does not belong in the cost at all
If the customer pays for design and fixtures separately — common for custom machinery — then those are not part of the product cost, and including them charges twice.
But settle ownership: if the fixture belongs to the customer and can move to another supplier, you cannot amortize it over future work either. You have neither the cost nor the security, and the sheet should reflect that.
Why this decides profit and loss
In a low-volume trade the non-recurring share is large. Omit it and the sheet looks healthy, the quote looks competitive, and you win the work — at a price below the real cost.
This is the mechanism behind a shop that is busy, quotes everything as profitable, and still cannot explain where the year went. Each job is profitable on the costs that were counted.
What to do when NRE is large
When the sheet shows a large one-time share, the improvement direction is not faster cycles. It is **making the one-time work reusable**:
- Standardize sub-assemblies across machine models, so one design and one fixture serve several orders.
- Use adjustable fixtures instead of dedicated ones wherever tolerances allow.
- Separate the reusable part of the design from the customer-specific part, and charge only the specific part to the customer.
None of these is visible while the one-time cost is buried inside a unit cost. Pulling it out is the precondition for seeing them at all.
Related guides
- Costing a machine build: roll up from parts, do not divide down from hours
A machine is an assembly of made parts, bought parts and subcontracted parts, each costed differently. How to roll them up, where the double count hides, and why commissioning is a cost line.
- Mold amortization and cavity economics: the arithmetic behind tooling decisions
More cavities means a more expensive tool and a cheaper part — up to a point. How to spread mold cost honestly, and how to compare two cavitation options with numbers instead of instinct.
- Six fields almost nobody fills in, and which way each one bends the cost
Material yield, personal and machine allowances, energy, units per pack and fixed cost per lot: six inputs that are usually left empty, and why five of the six make the cost look cheaper than it is.