Short answer
Work down the chain in a fixed order: the measurement, the sample, the light, and only then the process. Most apparent colour failures are not formulation faults. They are a drifted standard, a warm or unconditioned sample, a different instrument setting, a gloss difference or a metameric match, and each has a quick test that confirms or eliminates it. Adjusting the recipe before running those tests corrects a real batch towards a false reading.
The expensive mistake in colour control is correcting a batch that was right. If the standard has faded, the instrument settings have changed or the sample was read warm, a formula adjustment moves good material off target and the next batch, made to the adjusted recipe, fails for real. The checks are therefore ordered by cost and reversibility, not by how likely each cause feels: re-measuring costs minutes, re-presenting a sample costs an hour, a booth comparison under two lamps costs little more, and only a process investigation costs production time. The table lists twelve cases with the mechanism behind each and the test that isolates it. Its ranking of likely causes is an editorial judgement, not a measured frequency.
Start with the thing every reading depends on. Measure the working standard against the master or its recorded values; if it has moved, every recent result is suspect. Check the instrument against its check tile and confirm the settings match the specification: geometry, specular mode, aperture, illuminant and observer. Then measure the retained sample of the last good batch. If that now fails as well, the fault is in the measurement system and the new batch has not been shown to be wrong. When supplier and customer disagree, exchange the physical sample before exchanging opinions, because instruments of different models can differ on one piece by more than a tight tolerance allows.
A sample that is warm, damp, thin or badly presented gives a true reading of the wrong thing. Let it reach room temperature and re-measure; saturated reds, oranges and yellows are the most temperature-sensitive. Check opacity by adding layers or by measuring over black and over white. Re-present it several times and look at the spread: if repeat readings of one piece differ by a large share of the tolerance, the measurement cannot support the verdict either way. Compare gloss with the standard, and measure with the specular component included and excluded. Agreement with it included and disagreement with it excluded means the pigmentation matches and the surface does not, which is a finishing problem.
If the instrument passes the batch and people reject it, or it passes in the laboratory and fails at the customer's, take both pieces to the booth. View them touching, in the product's orientation, under the daylight source and then under the source that represents where the product is seen. A match that opens up under the second lamp is metameric: the batch reaches the colour with different colourants or on a different substrate, and the reflectance curves of the two will cross. A print that looks bluer than it measures points to optical brighteners and a mismatch between the ultraviolet content of the lamp and of the instrument. A difference that only some observers see can be observer metamerism, which no single instrument reading resolves.
When measurement, sample and light have been cleared, the direction of the difference indicates where to look. A batch that is weaker or stronger, differing mainly in lightness and chroma with hue on target, points to colourant strength, concentration or dispersion. A hue shift points to the balance between colourants, to a substituted lot, or to heat damage affecting one pigment more than the others. A difference confined to one tool, machine or shift points to process conditions. In every case the next step is the same: measure the incoming colourant lot against the retained previous lot before changing anything, since a formula fixes proportions by weight and not the strength of what was weighed, and then correct the batch from measurement.
| What you see | Where it sits | Mechanism | Test that isolates it | Most likely cause | Next action |
|---|---|---|---|---|---|
| A batch made to the same formula differs from the last one | Process | A formula fixes proportions, not the colour strength of what is weighed. Colourants vary from lot to lot, dispersion varies with mixing, and heat history changes some pigments. | Measure the incoming colourant lot at a fixed concentration against the retained previous lot before blaming the process. | A change of colourant lot or strength, then dispersion, then processing temperature | Check colourant strength on receipt and correct the batch from measurement, not by repeating the recipe. |
| The batch passes the instrumental tolerance but looks wrong | Specification | The number summarises one geometry, one illuminant and one formula. Gloss, texture, the direction of the difference and how close the parts sit all change what is seen without changing the figure. | View the pair in the booth, touching, in the product's real orientation; then read the lightness, chroma and hue components separately. | A tolerance shape that passes a hue shift, or an appearance difference the instrument is not measuring | Add component limits and a gloss tolerance, and re-derive the limit from samples the customer has accepted and rejected. |
| The match holds in the daylight booth and fails under shop or office lighting | Light and observer | The batch and standard reach the same colour with different colourants, so their reflectance curves cross; the two agree for one light and separate under another. | Switch the booth between its daylight and its shop-light sources, and compare the two reflectance curves for crossings. | Illuminant metamerism from a substituted colourant or a different substrate | Match with the standard's colourants where possible and add a metamerism limit under the illuminant the product is sold in. |
| Two parts in the same colour look lighter and darker beside each other | Sample | A glossier surface sends its surface reflection in one direction, away from the eye, and looks darker and more saturated; a matt one scatters it into the view and looks lighter and greyer. | Measure both with specular included and excluded. If they agree included and differ excluded, the pigmentation matches and the surface does not. | A gloss or texture difference, not a pigmentation difference | Tolerance gloss alongside colour, and fix the surface or tool finish before touching the formula. |
| The same ink or coating reads differently on a different substrate or lot of board | Sample | A film that does not fully hide takes part of its colour from what is beneath, and an absorbent or tinted substrate changes the film itself. | Draw the coating down over black and over white and compare the two readings; a gap means the film is not hiding. | Incomplete hiding or a change in substrate colour or absorbency | Approve the colour on the production substrate at production film weight, and specify the backing for measurement. |
| Moulded parts from one compound vary with tool, wall thickness or machine | Process | Mould texture changes surface scattering, a thin translucent wall lets the background through, and excess processing temperature degrades heat-sensitive colourants. | Mould the standard plaque on the same machine at the bottom and top of the temperature window, and measure a textured and a polished face of one shot. | Surface texture first, then thickness in translucent grades, then heat history | Fix the plaque tool, thickness and surface in the specification, and measure only parts cooled to room temperature. |
| A sprayed panel differs from the drawdown that was approved | Sample | Colour depends on film thickness until the film hides completely, and application method changes surface texture and, in effect finishes, flake orientation. | Measure the dry film thickness of both, and apply the batch by the approved method at the approved thickness over black and white. | A thin film that does not hide, or a different application method | Write film thickness, substrate and application method into the standard-preparation procedure. |
| A fabric reads differently each time it is measured, or panels cut from one roll look different | Sample | Yarn direction, nap and pile reflect light differently along and across the fabric, and a single thin layer lets the backing show. | Fold to opacity, then measure, remove, refold and re-measure several times; compare the spread with the tolerance. | Presentation: too few layers, a fixed orientation, or pile entering the port | Fix the number of layers, the aperture and the number of repositioned readings to average, and use a glass plate for pile. |
| A print matches its proof on one paper and not on another, or looks bluer in the booth than it measures | Light and observer | Optical brighteners in paper absorb ultraviolet and emit blue, so the paper's colour depends on how much ultraviolet the lamp or the instrument supplies. | Measure the unprinted stock in the M1 and M2 conditions; a large difference in b* means the paper is heavily brightened. | Different brightener levels between proofing and production stock, or mismatched measurement conditions | Specify the measurement condition and the paper, and judge the print under light whose ultraviolet content matches the instrument's. |
| Supplier and customer measure the same sample and get different results | Instrument | Instruments of different models, geometries or apertures make different systematic errors; even identical models drift apart between services. | Circulate one set of stable tiles and one real sample, and compare each site's readings with the first site's. | Different geometry, specular mode, aperture or illuminant and observer settings before any instrument fault | Align the settings, then the model; profile the fleet against shared tiles and keep the tolerance wider than the remaining disagreement. |
| A sample measured straight off the line fails, and passes an hour later | Sample | Many strongly coloured materials shift reversibly with temperature, and textiles shift with moisture content until they reach equilibrium with the room. | Measure the same piece at intervals as it cools or conditions and plot the readings against time. | Thermochromism or an unconditioned sample, not a batch fault | Set one measurement temperature and a conditioning time for standards and samples at every site. |
| Batches that used to pass now fail in the same direction, with no process change | Specification | The working standard has faded, soiled or been polished by handling, so every batch is being compared with a different colour from the one approved. | Measure the working standard against the archived master, or against the digital values recorded when it was approved. | A degraded working standard | Replace the working standard from the master, and record spectral values for every standard at issue. |
Each statement is labelled by kind — established fact, a standard’s requirement, observed market data, a convention, or Colourwise’s own interpretation or analysis — with the strength of the evidence behind it.
Colourwise interpretationModerate evidence
Checking the measurement system, the sample and the viewing conditions before the formulation avoids correcting a sound batch towards a false reading; the order is by cost and reversibility of the check.
Based on: Colourwise's synthesis of the standards and instrument makers' guidance cited across this section, each of which identifies a non-process source of apparent colour difference.
Caveat: A working method, not a published procedure; no data on how often each cause occurs is cited.
StandardStrong evidence
ASTM D2244-25 notes that where specimens may be metameric the instrumental result should be verified visually, and that gloss and texture can affect the correlation between measured colour difference and acceptability.
Source: ASTM D2244-25 Standard Practice for Calculation of Color Tolerances and Color Differences from Instrumentally Measured Color Coordinates; ASTM D4086-18(2023) Standard Practice for Visual Evaluation of Metamerism
FactModerate evidence
Instrument makers list mixed geometries or models, inconsistent calibration, variable sample presentation and aperture, and instrument drift as the sources of disagreement between colour measurements made at different sites.
Source: Why Benchtop Spectrophotometer Inter-Instrument Agreement & Repeatability Matter for Brand Suppliers
ConventionModerate evidence
Lot-to-lot variation in colourants is a recognised source of batch colour variation in plastics, and monitoring the strength of incoming pigments and dyes is recommended practice.
Source: Color and Appearance Measurement in Plastic Industry (white paper)
StandardStrong evidence
ISO 13655:2017 defines measurement conditions M1 and M2, with and without a defined ultraviolet component, so that the fluorescence of optically brightened papers is measured consistently with how the print is viewed.
Source: ISO 13655:2017 Graphic technology — Spectral measurement and colorimetric computation for graphic arts images; ISO 13655 measurement modes M0–M3
Reviewed 6 October 2026. Colourwise summarises its sources in its own words and does not reproduce standards text or proprietary colour data. Spotted an error? Tell us.