Home / Wiki / Quality & Inspection / Measurement Methods

CNC Part Measurement Methods

Every machined part must be verified before it ships. The question is not whether to measure, but how — and with what level of precision. A caliper that reads to 0.02mm is perfect for checking overall dimensions, but useless for verifying a bearing seat at ±0.01mm. This page helps you choose the right measurement tool for each tolerance level, understand what each method costs and how long it takes, and avoid the inspection mistakes that cause the most quality disputes between customers and suppliers.

Which Measurement Method Do You Need?

The tolerance on your drawing determines the minimum measurement method. Using a tool that is not accurate enough gives false confidence — the part reads "in tolerance" but is actually out of spec. The table below maps tolerance ranges to the appropriate measurement method. Always use a measuring instrument with at least 4–10 times the accuracy of the tolerance you are verifying (the 10:1 rule).

Tolerance RangeRecommended MethodAccuracy of InstrumentTypical Application
±0.1 mm and looser Digital caliper (0.01mm resolution) ±0.02–0.03 mm Overall dimensions, non-critical features, stock lengths, clearance holes. The workhorse of CNC inspection.
±0.05 mm Micrometer or digital caliper ±0.005–0.01 mm (micrometer) Fits, mounting faces, O-ring grooves, dowel holes. Micrometer preferred for critical dims; caliper acceptable for non-critical dims at this level.
±0.025 mm Micrometer, height gauge, or dial indicator ±0.002–0.005 mm Bearing journals, precision bores, sealing surfaces. Hand tools at the edge of their capability — CMM starts to make sense here.
±0.01 mm CMM or precision micrometer ±0.001–0.002 mm (CMM) Press fits, gauge features, precision tooling. CMM is strongly recommended. Hand tools can work for simple dims but are operator-dependent.
±0.005 mm and tighter CMM (temperature controlled) ±0.0005–0.001 mm Gauge blocks, optical mounts, semiconductor fixtures. Requires controlled environment (20°C ±1°C), skilled operator, and calibrated equipment.
GD&T (position, profile, runout) CMM ±0.001–0.002 mm Any drawing with GD&T callouts (true position, profile of a surface, circular runout, etc.). CMM is the only practical method for GD&T verification.
Surface roughness (Ra/Rz) Surface roughness tester (contact or optical) ±5–10% of reading Sealing surfaces, bearing surfaces, cosmetic finishes. Contact stylus for most applications; optical for soft materials or finished surfaces.
Small features (<1 mm), profiles Optical comparator or vision system ±0.001–0.005 mm Small radii, thin walls, edge break verification, profile comparison against overlay. Non-contact, so no risk of damaging delicate features.
The 10:1 rule Your measuring instrument should be 10 times more accurate than the tolerance you are verifying. For a ±0.05mm tolerance, the instrument should resolve to 0.005mm. For ±0.01mm, it should resolve to 0.001mm. If you are using a caliper (0.02mm accuracy) to check a ±0.05mm feature, you are at 2.5:1 — barely adequate. For anything tighter than ±0.025mm, switch to a micrometer or CMM.

Measurement Tools at a Glance

The table below summarizes every common measurement tool used in CNC machining inspection, with its accuracy, relative cost, what it measures, and when it is the right choice. This is your quick-reference for selecting inspection equipment.

ToolAccuracyRelative CostWhat It MeasuresWhen to Use
Digital caliper ±0.02–0.03 mm $ (20–200) External dimensions, internal dimensions, depth, step First-pass inspection, overall dimensions, non-critical features, incoming material checks. Every machinist has one.
Micrometer (outside) ±0.002–0.005 mm $ (50–500) External diameter, thickness, sheet metal Shaft diameters, pin diameters, thickness on flat parts, any external dimension that needs better accuracy than a caliper.
Micrometer (inside / bore) ±0.005–0.01 mm $ (100–800) Internal diameter, bore size Bore diameters, hole sizes, bearing seat bores. Three-point bore gauges are most common; bore micrometers for very high precision.
Height gauge ±0.01–0.02 mm $ (200–1,500) Height from surface plate, step heights, scribed layout Step measurements, feature heights from a reference surface, layout marking before machining.
Dial indicator / DTI ±0.005–0.01 mm $ (30–300) Runout, flatness, parallelism, deviation from reference Checking runout on turned parts, flatness of machined surfaces, alignment verification. Used on a surface plate or magnetic base.
Pin gauge set Fixed sizes (H7 tolerance) $ (50–500 per set) Hole diameter (go/no-go) Fast verification of hole sizes. GO pin enters, NO-GO pin does not. The fastest way to check hundreds of holes.
Thread gauge (go/no-go) Per thread standard (6H/6g) $ (20–200 per size) Thread pitch diameter (go/no-go) Checking internal and external threads. GO gauge threads in fully, NO-GO gauge does not enter more than 1–2 turns.
CMM ±0.001–0.002 mm $$$ (80k–500k machine + $30–80/hr operating) Any dimension, GD&T, 3D geometry Tight tolerances (<±0.025mm), GD&T verification, complex geometry, FAI, PPAP documentation. The gold standard for dimensional inspection.
Optical comparator ±0.005–0.025 mm $$ (10k–80k) 2D profile, radii, angles, edge quality Profile comparison against drawing overlay, small feature inspection, thread form verification, edge break measurement.
Vision measurement system ±0.001–0.005 mm $$$ (30k–200k) 2D dimensions, patterns, small features, optical edges Automated inspection of small parts, stamped parts, PCB features. Non-contact, high-speed, programmable for production.
Surface roughness tester ±5–10% of Ra reading $$ (2k–20k for portable; 20k–100k for benchtop) Ra, Rz, Rq, Rsm (surface roughness parameters) Verifying surface finish specifications on sealing surfaces, bearing surfaces, cosmetic parts.
Cost is per measurement, not per tool A $500,000 CMM costs nothing per part once it is installed. A $100 caliper costs $0.50 per part in labor time. The real cost of measurement is the operator time and the opportunity cost of tying up production equipment for inspection. For high-volume production, a $50,000 vision system that inspects a part in 3 seconds is cheaper per part than a $100 caliper that takes 2 minutes per part.

Calipers & Micrometers

Calipers and micrometers are the two most common hand-held measurement tools in CNC machining. Together, they handle the vast majority of dimensional verification for parts with tolerances of ±0.05mm and looser. Understanding when each is sufficient — and how to use them correctly — eliminates more inspection errors than any other knowledge.

Digital Calipers

A digital caliper measures external dimensions, internal dimensions, depth, and step height. It has a resolution of 0.01mm but an accuracy of roughly ±0.02–0.03mm. It is the most versatile hand tool and should be the first tool reached for on the inspection bench.

When a Caliper Is SufficientWhen You Need a Micrometer Instead
Tolerance ±0.1mm or looser Tolerance ±0.05mm or tighter
Overall dimensions (length, width, height) Mating diameters (shafts, bores)
Clearance hole diameters Press-fit and transition-fit diameters
Depth of pockets and holes Sheet metal and thin-wall thickness
Quick verification before CMM Features where 0.02mm error matters
Stock material verification Quality documentation (FAI, PPAP)

Micrometers

A micrometer measures external dimensions (outside micrometer) or internal dimensions (bore micrometer / three-point gauge) with an accuracy of ±0.002–0.005mm — roughly 5–10 times better than a caliper. It uses a ratchet stop or friction thimble to ensure consistent measuring pressure, which is the single biggest advantage over calipers.

FeatureDigital CaliperOutside Micrometer
Resolution 0.01 mm 0.001 mm (0.01 mm on some models)
Accuracy ±0.02–0.03 mm ±0.002–0.005 mm
Measuring pressure Operator-controlled (variable) Ratchet stop / friction thimble (consistent)
Measures External, internal, depth, step External (or internal with bore gauge)
Range per tool 0–150mm (typical), 0–300mm 0–25mm per frame (need multiple for larger range)
Best for General-purpose, non-critical dims Critical diameters, fits, thickness

Common Mistakes with Calipers & Micrometers

#MistakeEffectCorrect Practice
1 Too much measuring pressure on caliper Reading 0.02–0.05mm smaller than actual. The jaws flex under force. This is the #1 cause of caliper error. Use light, consistent pressure. The part should just barely slide between the jaws. Never force the caliper closed.
2 Not zeroing before use Systematic offset on every measurement. A caliper that reads 0.03mm at zero adds 0.03mm to every reading. Zero the caliper with jaws fully closed before each measurement session. Check zero periodically during use.
3 Measuring at an angle (not perpendicular) Reading larger than actual. The caliper jaw contacts at a point that is not the true diameter or length. Rock the caliper gently to find the smallest reading (external) or largest reading (internal). The true dimension is at the extreme.
4 Using a worn or damaged measuring face Inconsistent readings, especially on small features. Worn jaws give different results depending on where the part contacts. Inspect jaws for wear (light gap check against a flat surface). Replace or recalibrate when wear exceeds 0.01mm.
5 Using the wrong micrometer range A 25–50mm micrometer used on a 24mm part will give completely wrong readings. Each frame has a 25mm range for a reason. Always check the micrometer range matches the nominal dimension. Use 0–25mm for parts under 25mm, 25–50mm for 25–50mm, etc.
6 Measuring a hot part Thermal expansion causes the reading to be larger than the dimension at 20°C. Aluminum expands 0.024mm per 100mm per °C above 20°C. Let the part cool to ambient temperature before measuring. For tight tolerances, measure in a temperature-controlled environment.
The caliper confidence trap A digital caliper displays three decimal places (e.g., 25.123mm), which creates a false sense of precision. The display resolution is 0.01mm, but the accuracy is only ±0.02–0.03mm. That third digit is noise, not signal. Never trust a caliper reading to 0.01mm for a critical dimension. If the tolerance is ±0.05mm, the caliper margin is razor-thin. Use a micrometer.

CMM (Coordinate Measuring Machine)

A Coordinate Measuring Machine (CMM) uses a probe to measure the 3D coordinates of points on a part, then computes dimensions, distances, angles, and GD&T parameters from those points. It is the most versatile and accurate measurement tool available in a CNC machine shop, and the only practical method for verifying GD&T callouts.

When You Need CMM

CMM is not required for every part. Use this decision guide to determine when CMM inspection is justified.

SituationCMM Required?Why
Drawing has GD&T callouts Yes GD&T features (true position, profile, runout, perpendicularity, etc.) require 3D coordinate measurement. Hand tools cannot verify GD&T.
Tolerance ±0.025mm or tighter Yes (recommended) At this tolerance level, hand tools are at the edge of their capability. CMM eliminates operator variability and provides documented results.
First Article Inspection (FAI) Yes FAI requires documentation of every dimension. CMM generates the inspection report automatically.
PPAP / AS9102 documentation Yes Automotive (PPAP) and aerospace (AS9102) require CMM-generated dimensional data with statistical analysis.
Complex geometry (curves, contours) Yes (strongly recommended) Profile of a surface, complex curves, and 3D contours cannot be measured with hand tools. CMM or optical methods are required.
Customer requires CMM report Yes If the PO or drawing specifies CMM inspection, it is a contractual requirement.
High-volume production with SPC Recommended Statistical Process Control (SPC) requires consistent, repeatable measurement data. CMM provides this; hand tools introduce too much operator variation.
Tolerance ±0.1mm, simple geometry, no GD&T No Calipers and micrometers are sufficient and much faster. CMM would add cost with no benefit.
Prototype, 1–5 pieces, visual check sufficient No For quick prototypes where the customer will do their own verification, hand tools are adequate.

CMM Accuracy & Capability

Modern CNC CMMs (bridge-type, granite base) achieve accuracy of ±0.001–0.002mm over their full measuring volume. This is 10–20 times better than a micrometer and sufficient for virtually all CNC machining tolerances.

SpecificationTypical Value
Positional accuracy (MPEp) ±0.0015–0.003 mm (for a 400×600×500mm machine)
Repeatability ±0.001–0.002 mm
Probe types Touch-trigger (most common), scanning (continuous), laser (non-contact)
Software PC-DMIS, Calypso, PolyWorks, RationalDMIS
Typical measurement time per part 5–30 minutes (depends on number of features)
Programming time (first article) 30–120 minutes (one-time cost)
Operating environment 20°C ±1°C, low vibration, humidity control (for best accuracy)

CMM Cost Considerations

CMM inspection is quoted per part or per hour. Typical rates and what to expect:

Cost ComponentTypical RangeNotes
CMM programming (first article) $50–200 One-time cost. Amortized over the order quantity. For 100 pieces, this adds $0.50–2.00 per part.
CMM measurement per part $20–80 Depends on the number of features and GD&T callouts. A simple part with 10 dims costs less than a complex part with 50 GD&T callouts.
FAI report generation $100–500 Includes CMM programming, measurement, and full FAI documentation (AS9102 Form 1/2/3 or PPAP equivalent).
Fixture for CMM $200–2,000 Custom fixture to hold the part on the CMM table. Only needed for complex parts that cannot be held in standard fixturing.
CMM is an investment, not a luxury For orders with GD&T callouts, CMM inspection is not optional — it is the only way to verify the part meets the drawing. Requesting CMM inspection adds $20–80 per part, but it eliminates quality disputes, reduces the risk of shipping non-conforming parts, and provides documented proof that the part meets specification. In industries like aerospace, automotive, and medical, CMM inspection is a regulatory requirement, not a choice.

Optical & Vision Measurement

Optical measurement systems use light instead of physical contact to measure part dimensions. They are ideal for small features, delicate parts, and profile verification where a contact probe might damage the part or cannot reach the feature. The two main types are optical comparators (manual) and vision measurement systems (automated).

Optical Comparator

An optical comparator projects a magnified silhouette of the part onto a screen, where it can be compared against a drawing overlay or measured with screen crosshairs. It has been used in machine shops for decades and remains a cost-effective tool for 2D profile measurement.

FeatureSpecification
Typical magnification 10×, 20×, 50×, 100×
Accuracy ±0.005–0.025 mm (depends on magnification)
Best for 2D profile comparison, thread form, small radii, edge break verification, angle measurement
Limitations 2D only (cannot measure depth or Z-axis), operator-dependent, limited to features visible in silhouette

Vision Measurement System

A vision measurement system uses a high-resolution camera, motorized stages, and image analysis software to automatically measure 2D features. It is essentially an automated, high-accuracy version of an optical comparator.

FeatureSpecification
Accuracy ±0.001–0.005 mm
Measurement speed 1–30 seconds per part (programmed)
Best for Small parts, stamped parts, PCB features, pattern inspection, high-volume production
Advantages over optical comparator Automated, programmable, higher accuracy, generates digital reports, consistent results
Limitations 2D only, cannot measure internal features (blind holes, undercuts), surface reflectivity can affect accuracy

Limitations of Optical Methods

Optical measurement is powerful but has important limitations that are often overlooked:

LimitationDetailWorkaround
Reflective surfaces Shiny or polished surfaces scatter light and create false edges. The system cannot distinguish the true edge from a light reflection. Apply a thin coating (spray developer, talc), use polarized light, or switch to contact measurement (CMM).
2D only Optical methods measure projected profiles. They cannot measure depth, Z-height, or internal features (blind holes, deep bores). Use CMM for 3D features. Combine optical (for 2D profiles) with CMM (for 3D dimensions).
Edge definition Soft materials (plastics, rubber), chamfered edges, or burrs can create ambiguous edges. The system may measure the burr instead of the true edge. Deburr before measuring, use edge detection thresholds, or use contact methods.
Translucent/transparent materials Glass, clear plastics, and translucent polymers do not produce a clean silhouette. Apply opaque coating or use back-lighting with edge detection algorithms.
When to choose optical over CMM Optical methods excel when the part is small (<50mm), features are 2D, the part is delicate (thin walls, soft material), or the inspection volume is high and speed matters. CMM is better when the part has 3D geometry, GD&T callouts, or internal features. Many shops use both: optical for quick 2D checks and CMM for full 3D verification.

Surface Roughness Measurement

Surface roughness is measured separately from dimensional tolerance. It quantifies the microscopic peaks and valleys on a machined surface. The two most common parameters are Ra (arithmetic average roughness) and Rz (mean peak-to-valley height). Understanding the difference and knowing which to specify prevents both over-engineering and under-specifying surface finish.

Ra vs Rz

ParameterFull NameHow It Is CalculatedWhat It RepresentsTypical Use
Ra Arithmetic average roughness Average of absolute deviations from the mean line over the sampling length The "average height" of surface irregularities. Smooths out extreme peaks and valleys. Most common specification. Used on the vast majority of engineering drawings. Default for general machined surfaces.
Rz Mean peak-to-valley height Average of the 5 highest peaks and 5 deepest valleys over 5 sampling lengths The "extreme range" of surface irregularities. More sensitive to occasional deep scratches or high peaks. Used when a single deep scratch could cause a problem (sealing surfaces, fatigue-critical parts). Common in European and Japanese drawings.
Ra to Rz conversion (approximate) There is no exact conversion between Ra and Rz because they measure different aspects of the surface. However, a rough approximation is: Rz ≈ 4–7 × Ra. For a surface with Ra 1.6, expect Rz to be roughly 6.4–11.2. If a drawing specifies Rz, ask the customer for the Ra equivalent if your shop only measures Ra.

Contact vs Non-Contact Measurement

MethodHow It WorksAccuracyAdvantagesLimitations
Contact (stylus) A diamond-tipped stylus (2–5 μm tip radius) is drawn across the surface. A transducer converts vertical movement into an electrical signal. ±5–10% of reading Most widely accepted, well-standardized (ISO 4287), works on most materials, portable models available May scratch very soft materials (copper, aluminum foil). Cannot measure inside small holes. Stylus tip radius limits resolution on very fine surfaces.
Non-contact (optical) White light interferometry or confocal microscopy measures surface topography by analyzing reflected light patterns. ±3–5% of reading No contact (safe for soft, polished, or coated surfaces), measures 3D topography, very high resolution on smooth surfaces Expensive ($20k–100k), cannot measure very rough surfaces (Ra >10 μm), transparent/reflective surfaces require preparation

What Ra to Specify

Surface roughness should match the functional requirement. Specifying a smoother surface than necessary adds cost for no benefit.

Ra ValueVisual AppearanceTypical ProcessWhen to Specify
Ra 0.1–0.2 μm Mirror-like Lapping, polishing, super-finishing Optical reflectors, precision seals, medical implants. Very expensive. Only when absolutely required.
Ra 0.4–0.8 μm Smooth, visible machining marks only under close inspection Grinding, honing, fine turning Bearing surfaces, dynamic seals (O-rings, lip seals), hydraulic cylinder bores.
Ra 1.6 μm Smooth, fine machining marks visible Fine milling, finishing pass, reaming Fits (H7/g6), static gasket sealing surfaces, cosmetic visible surfaces. The most common "precision" specification.
Ra 3.2 μm Standard machining finish, visible tool marks Standard milling, turning, drilling General-purpose CNC parts. Non-sealing, non-bearing surfaces. The default finish for most CNC operations.
Ra 6.3 μm Rough machining marks clearly visible Roughing pass only Internal pockets, weight reduction features, non-visible surfaces. Minimum cost.

First Article Inspection (FAI)

A First Article Inspection (FAI) is a complete, detailed measurement of the first production part (or one of the first parts) against every dimension on the drawing. It proves that the manufacturing process produces parts that conform to the drawing before full production begins. FAI is mandatory in aerospace (AS9102), automotive (PPAP), and common in medical device and defense manufacturing.

What Is FAI?

FAI is not a quick check — it is a comprehensive verification of every feature, dimension, material, and process specified on the drawing. It is typically performed on the first part off the production machine (or the first part after any process change).

ComponentWhat It IncludesHow It Is Verified
Product FAI (Form 1 & 2) All part numbers, raw material specs, special processes, functional tests Material certificates, process records, test results
Characteristic accountability (Form 2) Every dimension on the drawing listed with its characteristic name, specification, and the process that produces it Engineering review of drawing vs manufacturing plan
Dimensional data (Form 3) Measured value for every characteristic listed on Form 2, with pass/fail determination CMM measurement, caliper/micrometer for simple dims, surface roughness tester for Ra, thread gauges for threads

When Is FAI Required?

TriggerFAI Required?Details
New part number Yes Every new part number requires a full FAI before production can proceed.
Design change (revision) Yes Any engineering change that affects form, fit, or function requires a new FAI on the revised feature(s).
Manufacturing process change Yes Change of machine, tooling, fixture, process sequence, or material source requires FAI on affected features.
Change of manufacturing location Yes Moving production to a different facility (even within the same company) requires a new FAI.
Production interruption (>2 years) Depends on customer Some customers require a new FAI if production has been dormant for 2+ years. Check the PO or quality agreement.
Repeat order (same process, same facility) No (if FAI is on file) If a valid FAI exists and nothing has changed, repeat orders do not require a new FAI. Verify the existing FAI covers all current drawing revisions.

AS9102 vs PPAP

The two most common FAI frameworks are AS9102 (aerospace) and PPAP (automotive). Both verify the same thing — that the part meets all drawing requirements — but use different documentation formats.

AspectAS9102 (Aerospace)PPAP (Automotive)
Industry Aerospace, defense Automotive, transportation
Standard SAE AS9102 (3 forms) AIAG PPAP (18 elements, typically levels 1–4)
Documentation Form 1 (part number accounting), Form 2 (characteristic accountability), Form 3 (dimensional results) PSW (Part Submission Warrant), DFMEA, PFMEA, Control Plan, MSA, SPC, dimensional report, material certs, etc.
Scope Focus on dimensional verification of the first part Broader: includes process analysis, capability studies, failure modes, and process control
Cost (to supplier) $200–1,500 per FAI $1,000–10,000+ per PPAP package (depends on level)
Lead time 3–10 business days 2–8 weeks (depends on PPAP level and complexity)
FAI is a customer requirement, not optional If the customer's PO specifies AS9102 or PPAP, the parts cannot ship without the completed FAI/PPAP package. Budget the time and cost into your quote from the beginning. A typical AS9102 FAI adds 3–10 days to lead time and $200–1,500 to the part cost. PPAP adds 2–8 weeks and $1,000–10,000. These are non-negotiable in the industries that require them.

Common Mistakes

These are the most frequent measurement and inspection mistakes we see in CNC machining — from both customers specifying requirements and shops performing inspection. Each one is avoidable.

#MistakeWhat HappensCorrect Approach
1 Using a caliper for ±0.01mm tolerances The caliper's accuracy (±0.02–0.03mm) is worse than the tolerance. Parts pass inspection but are actually out of spec. Quality dispute is inevitable. Use a micrometer or CMM for tolerances tighter than ±0.05mm. Follow the 10:1 rule: instrument accuracy should be 10x better than the tolerance.
2 Not specifying which dimensions to inspect The shop inspects nothing (or only overall dimensions). The customer receives parts with unchecked critical features. Quality failure discovered at assembly. On the drawing or PO, clearly list which dimensions require inspection. If all dimensions require inspection, state "full CMM inspection." If only critical dims, list them explicitly.
3 Measuring parts while still hot from machining Thermal expansion causes readings to be 0.01–0.05mm larger than the cold dimension. Parts pass inspection warm, fail cold. Particularly bad for aluminum (2.4x the expansion of steel). Always let parts cool to ambient temperature (20°C / 68°F) before final inspection. For tolerances tighter than ±0.025mm, use a temperature-controlled inspection room.
4 Specifying "CMM inspection" without listing features The shop writes a CMM program that measures 10 easy dimensions and misses the 5 critical ones. The report looks good but the critical features are unverified. List every dimension and GD&T callout that must be on the CMM report. Or attach a marked-up drawing highlighting the critical dimensions.
5 Using the wrong datum for GD&T inspection The CMM measures from a different surface than the drawing specifies. All positional and profile measurements are wrong. Parts pass inspection but fail assembly. Clearly define datum features (A, B, C) on the drawing. Ensure the CMM program establishes the same datum reference frame. Verify datum alignment on the first article.
6 Not zeroing or calibrating instruments Systematic error on every measurement. A micrometer that is 0.005mm out of calibration reads every dimension 0.005mm off. On a tight tolerance, this means every part is wrong. Zero instruments before each use. Calibrate on a regular schedule (annually for CMM, quarterly for micrometers, monthly for calipers). Keep calibration certificates on file.
7 Confusing Ra with Rz on the drawing Customer specifies Ra 1.6 but the shop measures Rz (which is 4–7x larger). The shop thinks the part passes; the customer measures Ra and rejects it. Always state the parameter explicitly: "Ra 1.6" or "Rz 6.3." Do not assume the other party knows which parameter you mean.
8 Requesting FAI/PPAP but not allowing time in the schedule The shop ships parts without completing FAI to meet the delivery date. The customer rejects the shipment because FAI documentation is missing. Everyone loses. Build FAI/PPAP time into the project schedule from the start. AS9102 FAI: add 3–10 days. PPAP: add 2–8 weeks. These times are industry standard.
9 Inspecting only the first piece, then assuming all are good Tool wear, thermal drift, and material variation cause dimensions to shift during production. Parts 1–10 are good; parts 50–100 are out of spec. Establish a sampling plan: first piece, last piece, and periodic in-process checks. For high-volume production, use SPC (Statistical Process Control) charts.
10 Not keeping inspection records When a quality issue arises months later, there is no data to trace the problem. The customer cannot verify what was inspected. The shop cannot prove the parts were good. Keep all inspection records (CMM reports, caliper logs, material certs) for the lifetime of the product or per contract requirements (typically 5–15 years for aerospace/automotive).
The single most important inspection principle Document everything. Every dimension measured, every instrument used, every calibration date, every material certificate. In a quality dispute, the party with the best records wins. If you measured it and recorded it, it happened. If you measured it and did not record it, it did not happen.