Whitepaper: The Total Cost of a PCB — Why the Quote Price Tells You 60% of the Story

In sixteen years of negotiating PCB and assembly contracts, the single most expensive mistake we see customers make is comparing quotes by per-board price. The per-board number is the most visible line on a quote sheet, and it's also the one that captures the least of what you actually pay. A serious cost model accounts for tooling amortisation, NRE, scrap allowance, rework reserve, freight, customs duty, financing cost, and the indirect cost of supplier-induced schedule risk. By our reckoning across 200+ programmes, the quoted per-board price typically represents 58–65% of the true landed cost per delivered, accepted board.

This whitepaper unpacks the other 35–42% line by line, presents the cost model we use internally, and provides a worked example based on a real automotive electronics programme. A downloadable spreadsheet template — Excel and Google Sheets — accompanies the published article so you can plug in your own quotes and compare apples to apples.

"The cheapest quote on a 10,000-unit programme cost us 23% more in landed cost than the second-cheapest, once we'd run the numbers through with three months of actual data. Quotes are sales documents; cost models are operations documents." — Pioneer Horizon Commercial Director

What the quoted price typically captures well

• Bare-board fabrication unit cost (driven by panel utilisation, layer count, technology features).
• SMT assembly unit cost (driven by component count, cycle time, machine utilisation).
• Standard test allocation (often a fixed per-board allowance).

What it typically captures poorly or not at all

• NRE (tooling, fixtures, programming) amortised over a volume assumption that may not hold.
• Realistic scrap and rework rates — quotes assume sample-build yields, not steady-state yields.
• Inbound and outbound freight, especially under volatile container-rate conditions.
• Customs duty on imported components — varies by HS code and shifts with policy.
• Inventory financing and working-capital exposure across a long supply chain.
• Indirect costs of supplier risk: missed dates, partial shipments, quality excursions.

Non-recurring engineering (NRE) and tooling are typically quoted as a one-time fee — stencil at $400, ICT fixture at $4,800, functional test development at $12,000, programming for the first AOI recipe at $2,400. Customers usually amortise these across the programme volume. The trap is that the quoted amortisation assumes the full volume gets built. When the volume drops — and 41% of programmes we've tracked across customers ship less than 70% of their original committed volume in the first 24 months — the per-board cost rises sharply.

Worked example

Take a programme with $24,000 of NRE and a quoted 10,000-unit volume:

• Quoted NRE amortisation: $2.40 per board.
• Actual volume of 6,500 units in 24 months (typical attrition): $3.69 per board — 54% above quote.
• Actual volume of 4,200 units (poor attrition): $5.71 per board — 138% above quote.

How we recommend modelling this

Use three scenarios: optimistic (committed volume), nominal (70% of committed), pessimistic (40% of committed). The pessimistic scenario is what you should price into commitments to your customer or internal forecast. The optimistic scenario is what your supplier is quoting. The gap between them is volume risk that someone is carrying — usually you.

NRE buy-back clauses

For larger programmes, we recommend negotiating an NRE buy-back schedule with the supplier: at agreed volume milestones (e.g. 25%, 50%, 75% of committed volume), the supplier rebates a portion of the NRE. This rebalances the volume risk and gives both sides skin in the game. We've executed this structure on six programmes; it works.

Hidden NRE

• Functional test fixture revisions (typical: 2–3 minor revs over a 3-year programme, $1,500–$4,000 each).
• BOM substitution requalification when a component goes obsolete (typical: 5–8 substitutions per 100 BOM lines over a 3-year programme).
• ECN absorption costs — every engineering change order is typically billed by the supplier at a labour rate; budget for 4–6 minor ECNs per year.
• Process re-validation if you move from one line to another inside the same supplier (e.g. moving to a higher-class lane).

The downloadable template includes a hidden-NRE provisioning line that reserves 8–12% of base NRE for these realistic over-runs.

These are the four categories where customer cost models typically diverge furthest from supplier quotes. Suppliers' quotes assume favourable conditions; landed cost reality includes the bad weeks too.

Scrap allowance

Most quotes assume a steady-state first-pass yield of 99.0–99.5%, which is realistic for a mature programme. They do not always include the prototype-and-ramp phase where yields can be 95–98%. The cost of those early-build boards has to be absorbed somewhere — typically priced into the unit cost as a "ramp scrap reserve" of 1–2%. We see customers omit this and end up over budget on first ramp.

Rework reserve

• Cosmetic-only rework (silk, masking): typically < $0.40/board, often absorbed by supplier.
• BGA reball or replacement: $35–85/board depending on package, with cycle-time impact.
• Connector resoldering (manual): $4–8/board for a typical connector.
• Net rework reserve we recommend modelling: 0.8–1.5% of unit cost for Class 2 work, 2.0–3.5% for Class 3.

Inbound freight on components

This varies wildly. Programmes where the customer supplies components (consigned model) effectively transfer the freight cost from the supplier line to the customer line — but the cost doesn't disappear. We typically see 3–6% of BOM value go to inbound freight + handling for high-mix electronics, with spikes to 10–14% during peak shipping rate cycles. Quotes often assume nominal rates from the calmer months.

Outbound freight

From a TN supplier to a North American or European customer, sea freight runs $0.20–$0.45 per board for typical industrial volume on a 40 ft container. Air freight runs 5–8× that. The realistic blend across a 3-year programme is sea-dominant with air spikes during recovery from any quality excursion — model it as 60% sea, 40% air for honest budgeting on time-sensitive programmes.

Customs duty

Highly programme-specific. HS code 8534 (printed circuit boards) attracts duty at varying rates by destination. For Indian-origin product entering the US under various preferential schemes, duty can be 0–4.5%. Entering Europe, 0–6.7%. Entering Brazil or Argentina, 16–22%. The duty rate changes; the template includes a settable line for your specific destination.

Customs and broker fees

• Customs broker fee: $80–250 per shipment in most jurisdictions.
• Bond fee (US): $0.5–2 per $1,000 of declared value.
• Documentation handling: $50–150 per shipment.

For high-frequency, small-shipment programmes (e.g. weekly air freight of 200 boards), broker and bond fees can add 1–2% to landed cost just on transactional overhead.

The categories in this section don't appear on any supplier quote, which is why customers under-budget them most consistently. They are nonetheless real numbers that show up on the customer's P&L.

Working capital exposure

From PO release on components to revenue recognition on finished goods, capital is tied up at every stage:

1. Component purchase: typically 30–60 days payment terms with distributors.
2. Component in transit / at supplier: 14–28 days dwell.
3. Assembly cycle: 2–8 weeks depending on programme.
4. Outbound transit: 3–6 weeks sea, 5–10 days air.
5. Customer incoming inspection: 1–4 weeks.
6. Customer payment terms: 30–90 days from invoice.

Across this chain, capital can be tied up 90–180 days. At a 10% cost-of-capital rate, that's 2.5–5% of programme value carried in financing cost. For programmes with very long supply chains and slow customer pay, this can exceed the supplier's margin on the programme.

Inventory financing for raw materials

Some programmes require strategic stock on long-lead-time components (MCUs, FPGAs). Carrying 12 months of safety stock on a $40 MCU at 10,000 units/year volume ties up $400,000 in inventory. At 10% cost of capital, $40,000/year in carrying cost — about $4 per assembled board.

Quality excursion cost

The hardest number to estimate but often the largest. A single 1,000-unit lot rejection costs roughly:

• Direct rework or scrap: $5,000–25,000 depending on stage.
• Air-freight replacement to meet customer schedule: $3,000–12,000.
• Engineering investigation labour: 40–120 hours at fully-loaded engineering rates.
• Customer relationship damage: not directly priceable but real.

If your supplier has a 0.5% lot rejection rate vs 0.1% for the next supplier, on a 50,000-unit annual programme that's roughly 4 vs 1 lot rejections per year — a $50,000–100,000 swing on direct costs alone before relationship effects.

The "managed cost" line

For internal modelling we include a managed-cost line at 2–4% of programme value to absorb the cost of programme management on the customer's side: engineer time spent in supplier reviews, schedule firefighting, ECN traffic, quality reviews. Most cost models ignore this; ignoring it makes lower-touch suppliers look more expensive than they are.

This whitepaper is anchored by a downloadable cost-model template — Excel and Google Sheets versions, both with the same formulas and the same scenario structure. The template is published alongside the article and is the artefact most of our readers reach for. Here is how to use it.

Inputs you need

• Supplier quote at unit, NRE, and tooling level.
• Committed programme volume and your honest assessment of nominal and pessimistic volumes.
• BOM cost breakdown — at minimum, total BOM value and the share of high-risk lines.
• Geographic origin and destination — for freight and duty look-ups.
• Payment terms with supplier and from your customer.
• Your cost of capital (most companies have an internal rate; if you don't, use 10% as a placeholder).
• Class of work (Class 2, Class 3) — drives rework reserve.

Three scenarios, side by side

The template produces side-by-side outputs for optimistic, nominal, and pessimistic scenarios. The pessimistic column is the one to plan against if the programme has any downside risk on volume. Most programmes do.

What you can read off the output

• True landed cost per accepted board, by scenario.
• Decomposition of cost into 12 categories (quoted unit, NRE, scrap, rework, inbound freight, outbound freight, duty, broker fees, financing, inventory carry, quality reserve, managed cost).
• Comparison view across up to 4 supplier quotes simultaneously — useful for RFQ evaluation.
• Sensitivity analysis: how the landed cost changes when volume drops to 70% / 40% of committed.

Real-world worked example

A page in the template is pre-populated with the case study of an automotive electronics programme we ran — 5,000 units, 12-layer Class 3 board, Indian-origin to a European customer. The quoted unit price was $42.10. The landed cost per accepted board across the full programme came in at $58.40, or 38.7% above quote. Every line item is broken out in the template so you can see where the extra $16.30 went. (For the engineering-and-yield story behind that programme, see our case study on the 12-layer EV BMS build.)

How to engage

The template is free; we ask only that you share back any improvements you make. If you'd like a one-hour walkthrough of how to apply the model to a specific programme you're evaluating, book a session with our commercial team — we run roughly twenty of these a year for prospective customers and existing partners.