Quality Economics • September 2026

Cost of Poor Quality in 3D Printing — Why Your Scrap Rate Undercounts the Real Loss | Precise3D

A print bureau tracked its scrap rate carefully and reported 4.1%. On that number it priced its work and agreed a service level with a customer. Nine months later the customer left, and the post-mortem found the real failure cost was between 11% and 14% of revenue. The scrap number was right — 4.1% of builds were thrown away. What it missed was the six hours of engineering time spent on each failed build, the reprint that consumed the same machine slot again, the rush freight on the replacement, and the two credit notes that never got coded as quality cost at all.

The Four Costs That Follow a Failed Build

Additive manufacturing has an unusually clean failure signal: the part either came off the plate or it did not. That clarity is deceptive, because the visible event is the least expensive part of the loss. The build material is often a small fraction of the total cost of a failure, particularly for engineering polymers where machine time and labour dominate.

CostTypical driverUsually tracked?
MaterialFilament, resin, supportYes — the only visible one
Machine timeOccupied slot, no outputRarely priced per hour
LabourDiagnosis, teardown, re-sliceAlmost never
RecoveryRush reprint, express freightCoded as logistics
ExternalCredit note, return, lost accountCoded as sales cost

The accounting failure is structural rather than careless. Machine time, labour and expedited freight are all real costs, but they land in different accounts from material scrap, so no single report ever shows the total. The only practical remedy is to build the roll-up deliberately: define a standard cost per machine hour, a standard rate for technical labour, and a rule that any expedited shipment caused by a quality escape is coded to quality, not to logistics.

The distinction between internal and external failure cost is the one that changes behaviour. Internal failure — caught before shipment — costs money. External failure — caught by the customer — costs money multiplied by a factor that depends entirely on how forgiving the customer is. A distributor or bureau whose external failure cost has never been calculated is pricing risk it has not measured.

Photograph of a quarantine area on a factory floor with a red-tagged tray holding several failed 3D printed engineering parts, a clipboard with a non-conformance form and a rejected-parts bin beside it

Building the Cost of Poor Quality Number

Cost of poor quality is calculated, not measured, because much of it consists of standard rates applied to countable events. The reliability of the number depends on whether the events are counted honestly, not on whether the rates are exact. A rough rate applied to a complete event log is far more useful than a precise rate applied to an incomplete one.

ComponentFormulaTypical input
Loss per failed buildmaterial + (hours × rate)Machine rate by class
Internal failureΣ losses caught pre-shipPer non-conformance log
External failurecredit + freight + labour + marginPer complaint
Inspection costhours × rateAppraisal labour
Prevention costhours × rateTraining, calibration, DOE

Two structural observations follow from this table. First, appraisal and prevention costs are quality costs too, and they belong in the same report as failure costs, because otherwise the report implies that all quality spending is waste. Second, the ratio between the four categories is the diagnostic. An operation whose spend is dominated by external failure is buying its quality at the most expensive possible point in the process; the same money spent on prevention would buy more.

Diagnostic Question: “What did quality cost you last quarter, all four categories?”
What you're looking for: An answer that only cites scrap material means the number is not being calculated, and the operation is probably making pricing decisions on a fraction of its real cost base. An answer that breaks out internal failure, external failure, appraisal and prevention means the operation can see where it is buying quality, and can redirect spend from failure to prevention with a measurable payback.

The 8D Structure and Why Steps Get Skipped

The 8D method is a disciplined sequence for resolving a quality escape permanently rather than repeatedly. Its value is less in its novelty than in its enforcement of order: each step is a gate, and the steps most commonly skipped are precisely the ones that prevent recurrence.

StepActionCommonly skipped?
D1Assemble cross-functional teamOften nominal only
D2Describe the problem with dataYes — described by opinion
D3Containment before root causeSometimes
D4Root cause, escape cause, systemic causeYes — only one cause sought
D5Permanent corrective actionRarely
D6Implement and validateRarely
D7Prevent recurrence systemicallyAlmost always
D8Recognise the teamAlmost always

The step that carries the most weight is D4, because a single root cause is usually insufficient. Three distinct causes have to be identified. The occurrence cause is why the defect was created. The escape cause is why it was not detected before it left. The systemic cause is why the management system allowed a defect to be created and escape. An operation that fixes only the occurrence cause will create the same class of defect again through a different mechanism, because the detection and system layers were never strengthened.

The step that most often gets abandoned is D7, and it is the step that determines whether the exercise had value. D7 asks what else in the operation has the same weakness. If a failure mode was missed at final inspection, D7 asks which other characteristics share that inspection method. This is the loop-closing discipline that an FMEA and control plan is built to institutionalise, and without it an 8D resolves one complaint and leaves the category open.

Photograph of a quality review meeting table in a 3D printing factory with a large printed part, dimensional inspection reports, a tablet showing a build log and a whiteboard with an eight-step process outline

Root Cause Analysis That Survives Contact with Reality

Root cause tools fail in print operations for a specific reason: the process is instrumented, so the data exists, but it is not consulted. A build log can show that the chamber temperature never reached setpoint before the first layer went down. That is a fact, and it should terminate the discussion. Far more often, the discussion continues on the basis of what the operator remembers.

  • Reconstruct the build from the log before forming a hypothesis. Chamber temperature, flow rate and any fault flags are recorded data. Start from them, and treat recollection as a last resort.
  • Test the cause, do not vote on it. A root cause that has not been demonstrated by turning it on and off is a hypothesis. The demonstration is usually cheap: rerun the failed geometry with the suspected factor restored to the correct value.
  • Look for the escape, not just the defect. A defect that reached the customer passed through every inspection step in the operation. Each pass-through is a separate finding.
  • Separate common cause from special cause. A defect that appears at a stable low rate across all work is a system problem, and reacting to individual occurrences of it makes the process worse. This is a statistical conclusion, not an intuitive one.

The statistical distinction is why an unstable process cannot be improved by exhortation. A process whose run-to-run variation has not been characterised will produce complaints that look random, and each will be treated as a special cause requiring a heroic response. The measurement discipline that makes the distinction possible is the same one covered in our reliability and uptime guide, and the parameter characterisation behind it is covered in our design of experiments guide.

Macro photograph of a printed engineering component with a dimensional inspection report and a failed counterpart beside it on a dark surface, showing visible layer separation on the rejected part

Deciding Where to Spend the Next Hour

Once cost of poor quality is calculated, the allocation decision becomes arithmetic rather than opinion. The general result across manufacturing is consistent and worth stating plainly: prevention spending has the highest return, appraisal spending has a moderate return, and internal failure spending has almost none because it resolves events rather than causes.

  • Where external failure dominates: the escape detection is broken. Spend on inspection method and on the controls that catch a defect before shipment, not on more reprinting capacity.
  • Where internal failure dominates: the process is not capable or not controlled at the point of creation. Spend on parameter characterisation and on control-plan adherence.
  • Where appraisal dominates: quality is being inspected in rather than built in. The expensive inspection step is usually compensating for a process whose variation is too wide.
  • Where prevention dominates and failure cost is low: the operation is in the desired state. Maintain it, and re-check that the input assumptions still hold.

The fourth case is worth guarding, because prevention spending is only valuable while the underlying failure modes are correctly ranked. A control plan or FMEA that has not been reviewed since a machine, material or geometry change is describing a process that no longer exists. The review-trigger discipline that keeps the ranking live is set out in our FMEA and control plan guide, and the evidence package customers in regulated supply chains will demand is described in our PPAP guide.

Diagnostic Question: “For your last customer complaint, which of the three causes did you fix — occurrence, escape, or system?”
What you're looking for: An answer naming one of the three, with a described corrective action and a date, means the operation has a functioning closed loop. An answer that describes only reprinting or apologising means the complaint was contained and the category left open.

Precise3D on Measured Quality Cost

At Precise3D, quality cost is a tracked production metric rather than a reporting afterthought. Our 3,500 sqm Shenzhen production network logs non-conformance events against a defined cost model that includes machine time and technical labour, not material alone, and reviews prevention versus failure spend on a fixed cadence. That is the same cost structure we hand to distributors, so a customer conversation about warranty or service levels starts from a shared definition of what a failure costs.

Our Pro X1 and OpenSource1 platforms pair a 320°C hotend with an actively controlled heated chamber and a rigid frame, which is what makes a documented control plan enforceable rather than aspirational — a chamber that holds setpoint removes one common cause of variation and stops it appearing in complaint data as an unexplained defect. Every unit ships with CE LVD (EN 62368-1:2014+A11:2017) and RoHS (EU 2015/863) documentation, and we supply distributors with build-log formats, spares lists and maintenance intervals so field failure data can be analysed rather than merely recorded.

Reviewed by the Precise3D quality and engineering team. Cost rates, ratios and framework guidance described here are industry-typical ranges provided for guidance and are not a guarantee of performance or savings for any particular operation. Validate against your own accounting policies, application requirements and regulatory obligations before implementation.

Flat lay photograph on a dark surface of a quality cost review desk with several printed engineering parts, a non-conformance form, a digital caliper and a calculator beside a printed cost summary sheet

Measuring Quality Cost?

Want the Cost of Poor Quality Worksheet and 8D Template?

Tell us what you print and where failures are showing up. We will send the cost roll-up worksheet we use internally, including the machine-hour and labour rate structure, plus an 8D form laid out for additive processes so the three root causes get treated separately.

← Back to Blog