The Decision That Sets Your Unit Cost

Two companies making the same product at the same volume can land 40% apart on cost of goods purely on how the thing is put together. Assembly labor is often the second largest line after components, and unlike commodity pricing it is entirely within your control.

The instinct of most first-time founders is backwards in both directions. Some hand-build far past the point where it makes sense. Others buy a robotic cell for a product that has not proven demand and end up with $180,000 of tooling for a design that changes six months later. The decision is arithmetic, and the arithmetic is not complicated.

What Manual Assembly Actually Costs

Hand assembly means a fixture, a torque driver, a written work instruction, and a person. In the US, fully loaded assembly labor including benefits, supervision, and facility overhead runs $28 to $45 per hour. In Mexico, $9 to $16. In Vietnam or coastal China, $5 to $12, though the gap has narrowed enough that the location decision now turns on the factors in manufacturing in China vs the USA rather than wage rate alone.

Convert that to per-unit cost with honest cycle times. A product taking 4.5 minutes of touch time at $34 per hour costs $2.55 in labor, plus 15% to 25% for line inefficiency, breaks, and rework. Call it $3.05. Multiply by annual volume: at 30,000 units that is $91,500 a year in assembly labor alone.

Manual assembly earns its place through three advantages that automation cannot match. Setup cost is near zero, usually a few thousand dollars in fixtures. Changeover is instant, so a design revision means rewriting a work instruction rather than reprogramming a cell. And a human catches anomalies no vision system was told to look for: a scratched housing, a wrong-color cable, a part that feels loose. For a first production batch, hand assembly is almost always the correct starting point, and running a proper pilot production run by hand is how you learn the true cycle time before committing capital.

What an Automated Line Costs

Automation covers a wide range and the price range is just as wide.

  • Single automated station such as a screwdriving cell, press-fit station, or dispensing robot: $25,000 to $70,000 installed.
  • Collaborative robot cell with gripper, fixture, and vision: $60,000 to $140,000.
  • Multi-station indexed line with feeders, transfer, and inline test: $250,000 to $1.2 million.
  • Full custom line for a high-volume consumer product: $1.5 million and up.

The quoted machine price is 55% to 70% of the true cost. Add integration and programming, part feeders which are usually the hardest element, custom nests, acceptance testing, installation, operator training, spares, and eight to eighteen months of lead time. Budget maintenance at 6% to 10% of capital per year and expect 85% to 95% uptime, not 100%.

Automated cycle times run 3 to 10 times faster than manual with far tighter variance, which shows up as a better process capability index when you track it with statistical process control.

The Break-Even Calculation

Run it as a straight annual comparison. Manual annual cost equals volume times touch time times loaded labor rate, plus rework cost. Automated annual cost equals capital divided by the depreciation period, plus maintenance, plus the reduced operator cost of one technician tending the cell, plus utilities.

A worked example. A product needs 6 minutes of touch time at $36 per hour loaded, so $3.60 per unit. Annual volume 80,000 units gives $288,000 in labor. An automated cell costs $420,000 installed, depreciated over four years at $105,000 a year, plus $34,000 maintenance, plus one technician at $78,000, plus utilities. Total roughly $222,000. Automation saves about $66,000 a year and pays back the capital in a little over six years of savings, which most boards will reject.

Flip one variable: at 200,000 units a year, manual labor becomes $720,000 while automated cost rises only modestly, and payback lands inside 18 months. That sensitivity is the whole story. Volume dominates, and a volume forecast you do not believe is not a basis for a capital purchase.

Three factors belong in the model beyond labor: quality cost, since a field defect can cost 20 to 100 times the labor saved; labor availability, because in many US regions the constraint is finding forty reliable assemblers rather than paying them; and product life, since a cell amortized over a product replaced in two years never pays back.

The Middle Path Most Products Should Take

The binary framing is the mistake. Almost every successful hardware company lands on selective automation: automate the two or three operations that are slow, ergonomically punishing, or quality-critical, and keep hands on everything else.

Good candidates for a single automated station are screwdriving with torque verification and logging, adhesive and gasket dispensing where bead consistency matters, ultrasonic welding, heat staking, leak and electrical test, and label application. Each of these is a discrete, repeatable operation with a measurable failure rate, and each can be bought as a semi-standard machine for $30,000 to $80,000 rather than engineered from scratch.

Poor candidates: flexible cable routing, packing irregular items, cosmetic inspection of complex surfaces, and any operation where the design is still changing.

Design Choices That Cut Both Costs

Before comparing labor rates, reduce the labor. Design for assembly routinely takes 30% to 50% out of touch time, and it makes automation cheaper too, because a product that is easy for a person is far easier for a robot.

The high-value moves: cut part count by combining functions into single molded pieces; replace screws with snap fits, since each screw costs 6 to 12 seconds installed; design for single-direction top-down stacking so no reorientation is needed; make parts self-locating with chamfers and pins; eliminate loose hardware such as washers and clips; make parts either obviously asymmetric or fully symmetric so they cannot go in wrong; and specify connectors that are keyed and audibly click.

A part count reduction from 34 to 19 pieces is a common outcome of one focused review, cutting assembly time, inventory lines, and defect opportunities at once. That work belongs in the broader value engineering pass, before any capital request.

Decide With Numbers, Not Fashion

Start manual, instrument the line so you know real touch times per operation, automate the station whose measured cost justifies it, and revisit the calculation each time volume doubles. The logic governing choice of manufacturing process by volume applies to assembly too: the right answer at 5,000 units a year is wrong at 500,000.

Model Your Line Before You Buy One

Projects House builds assembly cost models from real cycle-time data, runs design-for-assembly reviews that cut touch time before any capital is spent, and specifies semi-automated stations where the payback is genuine. Send your assembly sequence and annual volume through our contact form.