NPI Cost

NPI and Engineering Change Cost Estimation: Building a Defensible Quote

A cost breakdown for new product introduction and engineering change work: what drives the money, how to quote it, and where estimates go wrong.

New product introduction cost splits into two pools that estimators routinely blur: non-recurring engineering and recurring per-unit cost. NRE covers design, prototype tooling, pilot runs, and qualification, often 60,000 to 400,000 dollars for a mid-complexity mechanical assembly. Recurring cost is what each unit carries at volume. Quote them on separate lines. A buyer amortizing 250,000 dollars of NRE over a 50,000-unit first year adds 5.00 dollars per unit, but over 200,000 units it is only 1.25 dollars. The NPI Launch Cost calculator keeps the NRE pool and the amortization volume explicit so the per-unit loading is visible.

Per-unit cost is driven by material, direct labor, machine time, scrap, and overhead. On a typical machined-and-assembled product, material runs 45 to 60 percent of factory cost, direct labor 10 to 20 percent, and machine or line time 15 to 25 percent. Overhead applied as a burden multiplier of 1.35 to 1.65 on direct labor is where quotes quietly inflate or collapse. If your assembly labor is 2.75 dollars and your burden multiplier is 1.5, the loaded labor is 4.13 dollars. Estimators who forget to burden labor understate cost by a third before scrap is even added.

Scrap and yield are cost multipliers, not afterthoughts. If first-pass yield at launch is 88 percent, you consume 1 divided by 0.88, or 1.136 units of material and labor for every good unit shipped, an effective 13.6 percent cost uplift. Early production yields of 80 to 90 percent are normal for the first 90 days and climb toward 97 to 99 percent as the process matures. Quoting steady-state yield on launch volume is the single most common way NPI quotes come in low. Model the yield ramp explicitly across the first three months rather than assuming mature numbers on day one.

Engineering change cost is where programs bleed after launch, and it is almost always underquoted. A minor change averages 300 to 3,000 dollars, a moderate change 3,000 to 15,000 dollars, and a major change with requalification and inventory scrap can exceed 50,000 dollars. The hidden drivers are obsolete inventory write-off, requalification testing, and documentation cascade across drawings, work instructions, and control plans. Use the Engineering Change Cost calculator to itemize engineering hours, scrap, tooling rework, and requalification separately, because a lump-sum guess of a few thousand dollars misses the inventory and testing buckets that dominate expensive changes.

Build a defensible quote bottom-up, then sanity-check top-down. Bottom-up: sum material from the BOM, add loaded labor from DFA Assembly Time output, add machine time at the line rate, apply the yield uplift, then add burden and target margin. Top-down: compare the result against known cost per unit on a similar product per pound or per part count. If a comparable assembly runs 6.20 dollars and yours pencils to 4.10 dollars, you have likely missed a cost bucket. A 20 percent gap between the two methods is your cue to hunt for the omission before the quote goes out.

Prototype and pilot costs must be quoted at their real per-unit reality, which is 5 to 20 times production cost. A prototype that will cost 340 dollars each in a build of ten may cost 22 dollars at volume, because setup, soft tooling, and manual operations dominate low quantities. The Prototype Build Cost and Pilot Run Cost calculators separate one-time setup from per-unit build so you do not accidentally quote pilot economics to a production buyer, or promise production pricing on a pilot lot that has not amortized its tooling.

Labor loading for the launch team is a real cost line that estimators drop. A launch needs engineering, quality, and program management hours that the ECO Workload and Design Review Workload calculators can size. If a launch consumes 3 engineers for 4 months at a loaded 140,000 dollars per year, that is 140,000 loaded per engineer times 3 times one-third of a year, roughly 140,000 dollars of launch labor. Bury that in overhead and your recurring quote looks competitive while the program loses money. Carry launch labor in the NRE pool where it belongs.

Where estimates go wrong, the pattern repeats: mature yield assumed at launch, unburdened labor, NRE amortized over an optimistic volume, and engineering changes priced as a flat fee. Each error runs 10 to 40 percent low individually, and they stack. A quote that assumes 98 percent yield, forgets a 1.5x burden, and amortizes NRE over double the realistic first-year volume can understate true cost by more than half. Run the NPI Launch Cost and Launch Readiness Score calculators together so the budget reflects the ramp, not the steady state you will not reach for a year.

Published 2026-07-01.