It is important and common knowledge in the market that the mains aspects of developing a brewed coffee product are sensory quality and the ability to standardize the final brew. For this purpose, SCAA CUPPING evaluates 11 attribute points as described above.
Primarily, the importance of these attributes in consumer perception and their impact on choosing between a special or common product, for example, are understood. To support this, a consumer analysis published by the Associação Brasileira de Indústria de Café (ABIC) in 2021, which conducted a survey with 5,460 people and received 180 thousand answer in 14 cities across 9 different states, shows the distribution shown in Figure 6.
The interviewees were divided into groups based on their perception and distributed as follows:
- General Public: Common coffee lovers who do not actively search for or have in-depth Knowledge about the brewed coffee.
- Enthusiastic: Coffee lovers who have knowledge and actively seek information and are curious about the products and information relating to this subject.
- Experts: Coffee lovers who have formally studied and/or work directly with coffee products and brewing and have in-depth knowledge about the subject.
The results of the attributes contribution to brew coffee quality are described below (figure 6).
It is possible to understand that the consumer insight and the deciding factor for drinking brewed coffee is aroma, which is the main attraction and the first priority. In other words, aroma is the primary reason for choosing the product. The second factor is taste, representing another important aspect. All three groups, namely the general public, enthusiasts, and experts, have the same understanding when it comes to drinking brewed coffee.
The fact is that the sensory experience of brewed coffee is generally constructed through the olfactory epithelium, taste cortex, and visual cortex. In the other words, these senses are involved: the eyes, the nose, and the mouth. These components have receptors that receive input from the coffee in the human body.
The first contact is the eye, nose, and the brow contacting with the mouth, sequentially. Thus, the logic is that the local impact (visual civil structure) affects the consumer. However, this is the third attribute in importance. The consumer has the first contact with nose, and after that, the mouth, because the cup design embraces both the nose and mouth at the same time (Figure 7).
The impact occurs in the aroma because after the infusion of the ground coffee, hot water discharges flavonoids, phenolics, and antioxidant compounds. Consequently, the vapor comes into contact with receptors present in the nose and immediately activates the taste cortex. However, to fully experience the taste, it needs to occur through the mouth4,9,11,12,13.
These two points configure the true sensation of the drinking brewed coffee, which is the most important aspect. At this exact moment, coffee lovers either win or lose. This is the marketing vision – to sell the cup of the coffee, regardless of the extraction method or origin. It is understood that there is a long process until the cup is filled (Figure 8).
The green coffee is a basic material, and the fruit shows 228 different chemical compounds. After roasting, the same fruit has 971 substances. The answer to this is the presence of high-density compounds in green coffee which, during the roasting in the generation of new substances as shown in Table 114,15,16.
Table 1: General chemical compounds of green coffee and roasted coffee.
Chemical class
|
Coffee
|
Green
|
Roast
|
Hydrocarbon
|
41
|
74
|
Alcohol
|
24
|
20
|
Aldehydes
|
32
|
30
|
Ketone
|
21
|
73
|
Acids
|
3
|
25
|
Esters
|
26
|
31
|
Lactones
|
4
|
3
|
Phenols and Ethers
|
10
|
48
|
Furan
|
17
|
127
|
Thiophene
|
|
26
|
Pyrroles
|
|
71
|
Oxazole
|
|
35
|
Thiazoles
|
|
27
|
Nitrogen Compounds
|
38
|
208
|
Pyridines
|
|
19
|
Pyrazines
|
|
86
|
Amines and mixture of nitrogenous compounds
|
|
32
|
Sulfuric Compounds
|
7
|
47
|
Several
|
5
|
17
|
Total
|
228
|
791
|
Thus, the dataset of the Coffee Quality Institute (CQI) has 1338 samples, split into two genera: Arabica and Conilon, with 2.13% and 97.86% respectively. The samples were harvested from 2009 to 2018, and they come from 36 different countries. The countries with the most samples are Mexico (236), Colombia (183), Guatemala (181), and Brazil (132), presenting a total of 29 species.
In this first moment, the figures show six different attributes of sensory quality: Taste, Fragrance, Acid, Sweetness, and Balance. The Y-axis represents the value in points, which is the classification score of the brewed coffee. The X-axis is separated by: Body (Figure 9-A), Balance (Figure 9-B), Sweetness (Figure 9– C), Acidity (Figure 9– D), Aroma (Figure 9– E) and Flavor (Figure 9– F).
The results were separated by points classification of the green coffee and subdivided by the score attributed to each general criteria score 19.
The sweetness is part of the attribute inside the final grade (score), basically showing a score above 6 points on the scale of 0 to 10. However, the minimum score for the other attributes was around 6. This shows the equilibrium of the coffee samples evaluation.
Because of this equilibrium of the score, the challenge is to identify the impact of each attribute on the general score. Thus, a linear correlation for each attribute was made and is shown below (Figure 10).
The heatmap scale classifies no influence as 0, and strong influence or correlation as 1. Thus, sweetness does not directly impact the decision between a very good coffee or a nasty coffee. It is possible to verify that the general score has been most impacted by the improvement of taste and equilibrium, but sweetness has the minimum influence on the final score, based on the dataset applied to this study.
It is indicated that the point of similarity from the taster to the consumer is the taste because professionals seek equilibrium and consumers consider aroma when deciding to drink brewed coffee. The market commonly applies four mains post-harvest methods: washed/wet, natural/dry, pulped natural/honey, and semi-washed/semi-pulped. For better interpretation, the impact on the score composition was organized by post-harvest method on the y-axis and score on the x-axis (Figure 11).
The range of all samples between 80-85 points, regardless of the post-harvest method. It is possible to understand that the general score is not influenced by the applied method.
However, washed and naturally dried methods present the biggest outlier components in their group, but the reason for this is not precise. Thus, the marketing business traditionally shows the differences and specificities, as well as the uniqueness of the characteristics of brewed coffee by the country where it is produced. For this reason, a graph was created to show the behavior of the general score by country group (Figure 12).
The graph groups the origin country on the x-axis and the general score on the y-axis. Note that it is not separated by genus (Arabica and Conilon). The variation in scores remains between the range of 80-85. However, the biggest oscillation is in the data from Haiti and United States. Ethiopia has approximately 75% of the data with a mean score up to 85, and Brazilian coffee has the same percentage around a score of 82.
This case study includes coffee from 36 different origin countries, and for comparing the statistical significance between the samples shown in figure 12, a variant analysis (ANOVA) was applied. The ANOVA compares the averages of the groups and determines their statistical differences using the pandas library in the Python language.
The protocol of the test was developed for all the data and defined a comparison of the attributes between origin and scores. Thus, the ANOVA test compares all the data present in the dataset and validates the variation among the groups.
Normally, this analysis is done in other software where only 2 groups can be compared at a time, thus increasing the complexity of obtaining results. To solve this problem, a library present in the Python language was used, and a script was built for this dataset. Thus, it was possible to compare 36 countries with each other at the same time.
If the PR(>F) value is less than 0.05, it is statistically denoted that there is a difference in the scores of the analyzing countries. If this value is less, the samples have statistically significant differences. In this case, since PR(>F) is 0, statistically the origins have significant differences with each other.
The F factor is the variance of the each group divided by the variation of all groups, and this number is 4,33717,18. It means that the group averages varied within themselves.
The 75% of the data falls within the range of 81-83 in terms of scores, in the case of Brazilian coffee, the score variation is in the range of 84-87, with a minimum and maximum of 80-90. In others words, the variation of each country stays within the global range.