A defect can show up in more than one way

When inspecting green coffee, looking at bean color is a useful step, but not a sufficient one. A problem bean can show changes in color, shape, size, surface, and smell all at once. Relying on a single criterion risks missing cases that fall outside the typical pattern.

ISO 4149:2025 builds green-coffee inspection on both visual observation and olfactory assessment, aimed at identifying foreign matter and defects. That scope alone shows quality can't be read from a single feature.

For example, two beans might share a similar color, but one could have an unusual shape or a sign of damage on its surface. Conversely, a bean with a slightly different color isn't necessarily defective if that difference falls within the raw material's natural range.

So an inspector needs to look at a sample as a set of signals. Each signal raises or lowers the likelihood that a bean belongs in the group to be removed, but no single signal should be turned into an automatic conclusion.

Color enables quick detection, but has limits

Color is one of the easiest visual signals to use. Black, overly pale, or unusually colored beans are typically easy to spot when a sample is spread out under suitable lighting. That's also why optical sorting machines can efficiently handle many types of beans with a clear color difference.

But color is affected by lighting, background, and surface condition. The same two samples observed under different conditions can look different. Some defects also don't produce a large enough color change for the human eye, or a color-only system, to reliably detect.

The older ISO 4149:2005 standard once described assessing overall sample color and uniformity under defined lighting conditions; ISO 4149:2025 is now the version that has replaced it. The takeaway isn't that “color is useless” — it's that observation needs to be standardized and placed within a broader inspection process.

Color should be a strong screening signal, not the sole verdict.

Shape, size, and surface complement color

Shape can reveal a broken, deformed, or improperly developed bean. Size helps identify beans that fall outside a sorting group. Surface condition can show mechanical damage, insect activity, or other harm. All of this information can supplement color when assessing a sample.

What matters is not treating each criterion as an independent standalone ruler. A small bean isn't necessarily a defect; a bean with a different shape isn't necessarily one to remove. The meaning of a given trait depends on the batch's standard, the variety, the raw-material source, and the intended use.

In practical inspection, experienced inspectors usually don't ask “is this bean the right color?” and stop there. They look at the whole picture: is the color unusual, does the shape fit, is there anything suspicious on the surface, and does the sample's smell raise concerns?

That's also why inspection data should record the type of defect, rather than just a single “bad bean” percentage. Knowing which category an issue falls into gives a facility the ability to trace the cause and choose the right response.

Inspection criteria need to be tied to sampling and purpose

Even a good set of criteria can produce a wrong conclusion if the sample doesn't represent the batch. If an inspector only looks at a handful of nice-looking beans on top of a sack, the conclusion can't represent the whole batch.

ISO 4149:2025 is designed for inspecting green coffee to identify foreign matter and defects, serving both quality control and requirement-based inspection. This underscores the role of the method itself, not just the human eye.

For a production facility, inspection forms should separate out the necessary categories of information: foreign matter, defects, color or visual appearance, smell, and any batch-specific criteria. When an issue recurs, that data can be used to investigate the raw-material source, processing step, or storage conditions.

A coffee bean isn't a problem with a simple “pass” or “fail” answer based on color alone. Good inspection is a process of combining multiple signals with an appropriate sampling method, then making a decision based on a predefined standard.

References

• ISO 4149:2025 — https://www.iso.org/standard/87699.html