A CMM inspection report is not a pass/fail certificate by itself. It is a record of how a coordinate measuring machine sampled a part, how the part was aligned to a coordinate system, what nominal values the software compared against, and what actual results came back. Two reports that show the same dimensional numbers can carry very different levels of confidence depending on the alignment, the sampling, and the stated uncertainty. This guide explains how an engineer or buyer can read a coordinate measurement report in a practical order so they can accept, question, or reject a result with specific follow-up questions rather than a vague sense of doubt.
The aim is not to turn every buyer into a metrologist. It is to make the report’s assumptions visible, because those assumptions decide whether the headline numbers mean what they appear to mean.
What a CMM report actually contains
A coordinate measuring machine measures the geometry of a physical object by sensing discrete points on its surface with a probe, then reporting the probe position as a displacement in a three-dimensional Cartesian coordinate system with X, Y, and Z axes. Bridge-type machines move the probe along three orthogonal axes, and each axis has a position sensor; when the probe touches or otherwise detects a point, the machine samples those sensors and records the location and measurement vector. Repeating this produces a point cloud from which the software constructs features such as planes, cylinders, and circles.
A typical report therefore contains far more than a column of numbers. Expect to find: the part number and revision, the drawing and datum references, the probe tip and stylus configuration, the alignment or coordinate system definition, the measured features with nominal and actual values, and the deviations. Many reports also list the sampling strategy, the software version, and an uncertainty or accuracy statement. If a report omits the alignment, the sampling plan, or a traceability statement for the machine, treat the dimensional columns as provisional until those are clarified.
Alignment comes before any number
Alignment is the act of fitting the part into the coordinate system that the report will use. A CMM measures everything relative to the coordinate system established during alignment, so a changed alignment can change the reported deviation for the same physical part. Reputable programs document the alignment explicitly, including which features or datum targets were used and in what order.
When reading a report, ask what was used as the primary datum, the secondary datum, and the tertiary datum, and whether those match the drawing. If the report was aligned to a different set of surfaces than the drawing specifies, results may look good while the part is still out of tolerance in the intended reference frame, or the reverse. Datum definition is a design decision, and the measurement only makes sense inside that decision. A misaligned or loosely controlled alignment is one of the most common reasons two facilities disagree about the same part.
Nominal, actual, and deviation: reading the three columns
The core of most reports is a nominal, actual, deviation layout. Nominal is the design or basic value taken from the drawing or model. Actual is the value the CMM computed from the measured points. Deviation is the difference between them, usually actual minus nominal, and it is the number that gets compared against the tolerance band.
Three practical cautions apply. First, confirm the sign convention: a positive deviation may mean the part is oversized or may mean it is displaced in a positive direction, depending on the feature and the software settings. Second, check whether the reported actual is a single extreme point or a fitted feature value. For a hole, one report may quote the maximum material condition diameter while another quotes a least-squares fitted diameter; these are different quantities and must not be compared directly. Third, remember that a deviation inside the tolerance band is a statement about the sampled points, not proof about the entire surface. A deviation value is meaningful only together with the tolerance it is judged against and the feature definition behind it.
Position, profile, and true position in the report
Position tolerance, called simply Position in the relevant Geometric Dimensioning and Tolerancing framework, is defined as the total permissible variation that a feature can have from its true position. The true position is the exact coordinate defined by basic dimensions. A CMM report often expresses this as a diametrical or bilateral position value and may list a bonus tolerance when the position is evaluated at maximum material condition or least material condition rather than regardless of feature size.
When a position result is reported, verify which material condition modifier was applied, whether the value is diametrical or radial, and whether the basic dimensions used match the drawing. A position result quoted without its material condition and its diametrical convention can be off by a factor of two, which is the difference between a comfortable accept and a false rejection. The same discipline applies to profile, flatness, and other geometric results: the reported value is inseparable from the feature construction and the modifier under which it was evaluated.
Uncertainty is not the same as tolerance
Measurement uncertainty describes the range within which the true value of the measured quantity is expected to lie, given the measurement process and its known error sources. Tolerance describes how much variation the part is allowed to have. They are related but not interchangeable, and conflating them is a frequent source of argument.
The accuracy of a coordinate measuring machine is typically expressed as an uncertainty factor that varies with measurement distance, and it depends on factors such as probe repeatability and the accuracy of the linear scales. Probe repeatability for a touch probe can be on the order of one micron, but that figure is a probe characteristic, not the total uncertainty of a reported diameter or position on a real part. The task-specific uncertainty also depends on the feature, the sampling, the surface finish, the alignment, and the part geometry. A single quoted machine accuracy specification, offered without the task context and without a traceability statement, is not a substitute for a task-relevant uncertainty budget.
For conformance decisions, the practical question is whether the uncertainty was considered when judging against the tolerance. Under decision-rule approaches, a result near the tolerance limit may be declared conforming, non-conforming, or indeterminate depending on the agreed rule and the size of the uncertainty relative to the tolerance. If a report claims a borderline feature passes, ask what uncertainty was attributed and what decision rule was used. Numbers are being reported with more digits than the process can justify if no uncertainty was ever computed.
Sampling: how many points, and where
Sampling is the choice of how many points are measured and where they sit on each feature. It has a direct effect on what the report can and cannot detect. A diameter measured from a few points on a lobed or out-of-round bore can read very differently from a diameter measured from a dense scan, because discrete probing samples the surface rather than describing all of it.
When reviewing sampling, look for the number of points or the scan density per feature, the distribution around the feature, whether a form error such as roundness was assessed, and whether the sampling matches the drawing requirement. A drawing that controls roundness or cylindricity cannot be validated by four points on a diameter. Conversely, dense scanning of a soft or thin-walled part can itself deflect the surface and bias the result, so sampling strategy and part stiffness have to be considered together. Ask whether the sampling plan was fixed in advance or chosen after seeing early results; the latter undermines the objectivity of the report.
A practical report review checklist
Use this report review checklist in order, and stop to ask when any item is missing.
- Identity and revision. Do the part number, serial or lot, drawing revision, and program revision on the report match the parts actually under review?
- Traceability. Is the CMM identified, and is there a calibration or traceability statement establishing that the machine’s accuracy was verified against traceable standards?
- Alignment. Are the primary, secondary, and tertiary datums stated and consistent with the drawing?
- Nominal source. Is the nominal taken from the correct drawing revision or model, and are basic dimensions used for position correctly?
- Nominal actual deviation. Check the sign convention, the feature-fitting method, and the units, and confirm the deviation is judged against the specified tolerance.
- Feature definitions. For holes and slots, confirm whether values are maximum material condition, least material condition, or least-squares fitted, and whether position values are diametrical or radial.
- Uncertainty. Is a task-relevant uncertainty stated, and was a decision rule applied for features near the tolerance limit?
- Sampling. Are point counts or scan density, distribution, and form assessment adequate for the tolerances being verified?
- Out-of-tolerance handling. Are failed features flagged with their magnitude, and is the number of parts measured stated?
- Environment and setup. Is the part fixtured without distortion, and is the reported temperature context appropriate for the stated tolerances?
Turning the review into useful questions
Once the checklist is applied, the follow-up questions become specific rather than general. Ask which datum features drove the alignment and whether they match the drawing. Ask what uncertainty was assigned to a borderline result and under what decision rule it was judged. Ask how many points were taken on a flagged feature and whether a form check was performed. Ask whether a reported diameter is a fitted value or an extreme value, and whether a position value is diametrical or radial. These questions are constructive: they let a supplier explain the reasoning behind a number, and they let a buyer distinguish a genuine capability problem from a reporting or convention mismatch.
The deepest point is that a CMM inspection report is an argument, not just a dataset. It argues that a specific alignment, a specific sampling plan, and a specific uncertainty treatment together support the claim that the part meets its tolerances. Reading the report well means reading that argument, and knowing when to ask for the missing premises before acting on the conclusion.
Related reading on this site: CNC Machining Quality Inspection: Buyer Guide to Reports and Checks, First Article Inspection for CNC Parts: A Buyer Checklist, Datum Schemes for CNC Parts: A Practical RFQ Guide, and CNC Tolerance Stack-Up: How Buyers Can Prevent Assembly Fit Problems.
