Measuring coffee roast color by Agtron: Why do whole beans and ground coffee yield different results?

A freshly roasted coffee batch may appear fairly uniform to the naked eye, but when fed into an Agtron machine, the results between the whole bean and ground coffee samples can sometimes differ significantly.

This is not necessarily a device malfunction. Coffee beans are not a homogeneous material from surface to core. During roasting, heat is transmitted from the outside in, and chemical reactions occur at different rates, and the degree of change in each region of the bean is not entirely the same.

Therefore, the color after roasting is not just a sensory characteristic. When measured by a suitable device and standardized process, it becomes a useful index to control consistency between batches.

1. What does Agtron actually measure?

Agtron is the commonly used term in the coffee roasting industry to refer to the system for measuring roast level through the energy reflection characteristics of the coffee sample.

The Agtron E20-CP III device is described by the manufacturer as a compact spectrometer designed specifically to evaluate both whole coffee beans and ground coffee. The device uses the near-infrared (NIR) region, rather than relying solely on visual color perception.

The basic principle is to shine energy onto the sample, receive the reflected signal, and use a standardized system to convert the signal into a corresponding index corresponding to the roast level.

As coffee is roasted to a deeper level, the chemical and structural changes in the bean alter the reflection characteristics. On the commonly used Agtron scale, lower readings typically correspond to darker roasted samples, while higher readings correspond to lighter roasted samples.

Important note: Agtron is not simply a camera that takes a picture and compares the brownness of the bean. It measures a standardized optical characteristic specific to the device and measurement scale.

2. Are Gourmet Scale and Commercial Scale the same?

Agtron has two commonly used measurement scales: Gourmet Scale and Commercial Scale.

These two scales cannot be arbitrarily substituted for each other. The Agtron E20-CP III device can be calibrated to one of the two scales.

For example, a roaster announces a product with an Agtron index of 60 but does not specify whether it is Gourmet or Commercial, this information is not sufficient for other roasters to replicate the results.

That is why quality control data must be clearly stated, including the model of the device, the measurement scale, and the type of sample.

9. Why do green beans and ground coffee have different Agtron numbers?

This is the most important point when building a quality control process after roasting.

Measure on green beans

When measuring green beans, the device mainly receives the reflection signal from the surface of the bean. This surface may show changes due to the roasting process, but it does not fully represent the degree of change inside the bean.

If the roast batch has a temperature difference between the surface and the core, the green bean measurement may not fully represent the differences.

Measure on ground coffee

When grinding the beans, the inside is exposed. The ground coffee sample is then distributed into a measuring layer, allowing the device to receive a signal from a more uniform collection of materials than the surface of each individual green bean.

This can significantly change the measured value.

A 2022 study by Yeager and colleagues published in the Journal of Food Science found that the Gourmet Agtron index and L*a*b* color values of the ground coffee sample can be higher than those of the green bean sample. The authors explained that the grinding process makes the sample more uniform and exposes the less roasted interior.

Do not assume that the difference between green beans and ground coffee is always fixed. It depends on the variety, bean structure, roast level, uniformity, and sample preparation method.

4. What does the SCA's Agtron 63 benchmark mean?

In the guide for the Q Grader exam preparation, the Specialty Coffee Association (SCA) provides a reference benchmark for coffee used in the roasting and cupping practice session:

SCA also illustrates cases of over-roasted, under-roasted, roasted too quickly, and roasted too slowly samples. Notably, two roast batches can have the same color index but have very different roast times.

That is why color roast cannot replace temperature profile evaluation and sensory assessment.

5. What can the roast color tell us about the flavor?

During the roasting process, Maillard reactions, sugar breakdown, chlorogenic acid transformation, and other heat-related reactions form compounds that affect color and flavor.

An increase in browning usually goes hand-in-hand with changes in the chemical structure of the bean. However, color is only a general indicator and cannot accurately determine the extent of the formation or breakdown of specific compounds.

In 2025, Anokye-Bempah and colleagues published a study on Scientific Reports investigating seven roast profiles with the same total roasting time but different heat intensities, using three coffee origins in a 5 kg commercial drum roaster.

The research team found that color values in the L*a*b* color space can be described by a common roast color curve at critical development milestones. This is a meaningful area of research for standardizing color measurement in roasting, but it does not mean that all roasting machines, coffee varieties, and production conditions can use a single color model without re-verification.

6. L*a*b* and Agtron: Two different representations

In addition to Agtron, the food industry also uses the CIE L*a*b* color system to describe color with three coordinates:

• L*: lightness, from dark to light.
• a*: green to red axis.
• b*: blue to yellow axis.

The two systems cannot be converted arbitrarily by a simple formula. Agtron is an index built on a separate measurement system of the device, while L*a*b* describes color in a standard color space.

In 2024, Carvalho Pires and colleagues evaluated three color acquisition methods, including a desktop colorimeter, smartphone image, and color sensor. The study used 70 roasted coffee samples, including both green and ground coffee.

With the desktop colorimeter, the best model achieved R² = 0.99 for both sample groups; the reported RMSE error was 1.91% for green coffee and 0.87% for ground coffee.

The results showed that the low-cost measurement method has the potential to support roasting control when a suitable calibration model is developed. However, the results of a research sample set do not demonstrate that every smartphone or camera app can replace a dedicated colorimeter.

7. Building a roast color measurement process for a roaster

For a roaster that needs to control uniformity between multiple production batches, SAVINA recommends the following internal procedure for standardization:

Step 1: Determine the measurement scale

Choose Gourmet or Commercial and keep it consistent throughout the monitoring system. If the measurement scale is changed, re-calibrate according to the manufacturer's instructions.

Step 2: Standardize the sample

• Take a representative sample from the roasted batch after cooling.
• Do not measure the sample while it is still hot, as temperature can affect the results.
• If measuring green coffee, use the same sample arrangement.
• If measuring ground coffee, keep the grinding equipment, fineness, leveling, and tray filling stable.

Step 3: Check the device calibration

Perform calibration using a standard disc or method specified by the manufacturer. Do not use the results of an unverified device as a standard for the entire production line.

Step 4: Repeatability Measurement

Measure multiple times on a representative sample, record the average result and the degree of variation. This is a proposed internal control procedure to evaluate the repeatability of the measurement, not a mandatory number specified by SCA.

Step 5: Comparison with Roast Profile and Cupping

Compare the roast color with the data:

• BT – Bean Temperature.
• ET – Environmental Temperature.
• RoR – Rate of Rise.
• Development Time Ratio.
• Weight Loss – weight loss ratio.
• Cupping results after the coffee has rested properly.

Refer to SAVINA's in-depth articles on BT, ET, and RoR in coffee roasting, DTR and development time, as well as Maillard reaction in coffee roasting.

The value of color measurement in coffee production

Agtron is not a tool for determining whether coffee is good or bad. Its actual value lies in its ability to convert a relative statement like 'this batch is roasted darker than the previous one' into data that can be tracked and compared.

When combined with proper sample preparation, calibration, heat profile control, and sensory evaluation, color measurement becomes a crucial link in the quality management system of a roaster.

For a professional roaster, what's worth considering is not achieving a beautiful Agtron number, but the ability to reproduce the desired product characteristics across multiple roasting batches.

References

1. Agtron Inc. – E20-CP III Coffee Roast Analyzer. Specifications and operating principle.
2. Specialty Coffee Association – How to Prepare for the Q Grader Course. Roast level and brew temperature standards for Agtron 63 ground and approximately 58 whole bean.
3. Yeager et al. (2022). Roast level and brew temperature significantly affect the color of brewed coffee. Journal of Food Science. DOI: 10.1111/1750-3841.16089.
4. Carvalho Pires et al. (2024). Feasibility of using colorimetric devices for whole and ground coffee roasting degrees prediction. Journal of the Science of Food and Agriculture. DOI: 10.1002/jsfa.13376.
5. Anokye-Bempah et al. (2025). A universal color curve for roasted arabica coffee. Scientific Reports. PMID: 40624084.
6. Bicho et al. (2012). Use of colour parameters for roasted coffee assessment. Food Science and Technology. DOI: 10.1590/S0101-20612012005000068