There is no universal “cost per print.” There is only a cost model with assumptions that fit a particular printer, workshop, product, and delivery standard. The goal is not false precision; it is to make every important assumption visible.
1. Start with material consumed, not model weight alone
For a basic FDM job, material cost is filament price per kilogram multiplied by the grams reported by the slicer. A 120 g part from a $24 spool uses $2.88 of model material. Include supports, purge material, skirts, brims, and color-change waste because they leave the spool even when they do not remain in the finished product.
If you regularly discard the last portion of a spool because of moisture, tangles, or color inconsistency, reduce the usable spool weight in your own model. A $24 spool that reliably yields only 900 g of saleable work effectively costs $26.67 per usable kilogram.
2. Measure electricity at the wall when it matters
Electricity equals average power in kilowatts multiplied by print hours and your energy price per kilowatt-hour. Do not multiply the printer’s maximum rated wattage by the whole job. Heated beds and hotends cycle, so average consumption is often much lower than the label rating.
A plug-in energy meter gives the best workshop-specific value. Measure a few representative jobs: a small PLA print, a large bed-filling print, and any high-temperature material you sell. Use a weighted average or keep separate presets.
3. Give the machine a cost per productive hour
Printers do not last forever and upgrades do not eliminate depreciation. A simple planning method divides purchase price by expected productive service hours. A $650 machine expected to deliver 5,000 print hours contributes $0.13 for every active hour.
This is not accounting depreciation. It is a replacement reserve: a way to ensure today’s orders help fund tomorrow’s machine. Recurring nozzles, beds, belts, lubricant, and fans can be added as a separate hourly maintenance allowance when your records are good enough.
4. Charge hands-on labor, not autonomous machine time
An eight-hour print does not require eight hours of operator labor. Record the minutes spent reviewing the file, slicing, loading material, preparing the bed, starting the job, removing supports, cleaning, sanding, assembling, inspecting, and packing. Multiply those minutes by a defensible hourly rate.
Repeat products should become faster. Time several batches and divide shared setup and packing time by the number of saleable units. That prevents a one-off workflow from making mature batch work look artificially expensive.
5. Treat failures as expected cost
If direct cost is $10 and the failure rate is 10%, adding $1 looks reasonable but slightly understates the expected cost. You have a 90% success probability, so the cost required to obtain one successful unit is $10 divided by 0.90, or $11.11.
Use observed failures by product family when possible. Tall multi-color prints and simple flat PLA parts do not deserve the same allowance. Track failures that cause a full reprint separately from cosmetic seconds you can still sell.
6. Add order-level and business overhead carefully
Packaging, labels, inserts, adhesive, and seller-paid postage can be attached directly to an order. Rent, software, web hosting, accounting, tools, and product photography are overhead. Divide overhead by realistic monthly saleable units or allocate it as a percentage of revenue—then review the assumption as volume changes.
Keep cost and price separate. Cost tells you what the product consumes. Price must also cover marketplace fees, taxes where applicable, customer support, discounts, and profit. A clean cost model gives you a floor; customers and positioning determine the ceiling.
Use the model