Sensory Attribute Specifications and Quality Control
Sensory specifications are more difficult to define than more literal physical measurements. It involves development of a common language and reasonable agreement as to the intensity of the experience. One individual’s perception of “high acidity” may be another individual’s perception of “medium acidity”.
Agreement as to relative intensity of various flavor attributes requires calibration among panelists as to language that will be used to qualitatively describe the flavor attribute and the relative scale put into place so that panelists can describe the intensity of that flavor. Panelists have different levels of sensitivity, experience, and innate preferences, however; so strict panelist agreement even with a calibrated panel is unlikely. This is the point of using a panel: to determine the response to a product from a reasonably wide variety of judges.
Since the results of sensory tests can become overly subjective, standards must be carefully set and adhered to. Procedures of testing must be consistent and product specifications well understood. In this paper, procedures of sensory testing are first put forward, including the handling of samples and submission to panelists. Then an example of how specifications, forms, and charts can be used to bring more consistency into sensory testing programs.
Sensory Testing Procedures
The sensory analysis process known as “cupping” seeks to use the human sensory system as a measurement device. Since an individual’s sensory system can be arbitrarily and unexpectedly affected physically (such as by illness) and psychologically (such as moods), results of sensory testing will always be somewhat subjective unless strict procedures are followed. To make these results as objective as possible, the following is suggested:
A panel of tasters, rather than a single taster, ensures a set of independent results. Even in small roasters where one individual is responsible for most product decisions, a panel of tasters should occasionally be convened or other tasters should be invited to evaluate the product. Including highly motivated but less experienced tasters on a panel can often provide a new perspective.
The panelists should be trained using previously tested samples and regularly evaluated in terms of their ability to appropriately discriminate flavor attributes. A cupper must be constantly sharpening their skills and abilities, not only in coffee tasting but other beverages as well.
All samples should be disguised using 3-digit codes. Numerous studies show that this convention helps ensure objective results (tasters, even professional ones, seem to have favorite numbers). Certain products can be cupped together (perhaps several Sumatras or other natural processed coffees), but the flavor of the sample should be its main identifying trait at the cupping table.
Samples should be uniformly prepared according to well established standards. These standards include setting up a clean and quiet environment with a minimum of distractions in which cuppers can concentrate. Suggested are the standards developed by the SCAA technical standards committee, which has taken current sensory practices and applied tem to traditional cupping. These standards include water quality, water temperature, grind, coffee-to-water ratio, and procedures.
Use of standardized forms, the simpler the better. One wants the cuppers to concentrate on the samples, not get hung up filling out forms.
The setting up of a test also requires proper record keeping and chain-of-custody of samples. Chain-of-custody should include what the sample is, where it was collected, and the time at which it was collected. Record keeping should have all this information along with the results of the test.
Development of Specifications for Individual Coffees
Specialty coffee roasters offer a wider variety of products than most commercial roasters and the differences between the products offered have more to do with different flavor profiles of products than different price levels. A coffee that appeals to one group of consumers may not appeal to others, though some coffees have a broader appeal than others.
Since these coffees are meant to be different, it makes sense to develop different standards for each product. The specifications should indicate in reasonable detail the most noticeable flavor attributes of each product and their intensities.
Free of defects: A consumer may prefer one coffee flavor over another, but virtually all consumers will reject coffee that is unpalatable. At the top of the list of any set of specifications is “free of defects or off-flavors”. A defect is defined here as flavor or physical condition that renders the product unsuitable for consumption and/or will cause immediate sensory rejection of the product (the “yuck” response). In terms of coffee, examples include defects originating in green coffee, such as ferment, phenol taste, or dirtiness that will survive any degree of roast. Off-flavors (also called “taints”) are similar but less intense.
Two simple sensory tests can reveal potential defects and how regularly they will occur in a sample. One is to smell the green coffee; many sensory defects will be obvious before the coffee is roasted. Some beans will stand out as black, red, or white and a visual observation of a random sample can also reveal if defects are present. A quick count of the most obvious variations can give an indication that a sample is defective.
Defects are standards of what should not be present in a product. However, coffee is a natural product and an occasional defective cup will appear in the best samples. If a defect is found, more testing should be performed to indicate if the problem is widespread or simply an isolated anomaly. An entire batch of coffee, in roast or green form, should not be rejected due to one bad cup (or one black bean) if it is reasonable to assume that the sample does not indicate overall quality. To determine if this is the case, another random sample should be taken and another set of tests performed. Of course, this represents a greater expense and should be done only if necessary and other coffee is not available.
A particular defect of great concern to the coffee industry is the defect known as “phenol”. The origin and level of occurrence of this defect cannot, at this writing to the best knowledge of this author, be statistically predicted on the basis of standard quality control practices. This is the sole exception to the “single bad cup” rule stated in the previous paragraph; it is reasonable to reject a batch due to a single instance of phenol.
Sensory Standards: The flavor attributes of the coffee are specified both qualitatively (a verbal description of perceptions) and quantitatively (the intensity of the perception). Taken together, these specifications will result in a coffee that cuppers within the roasting company will clearly identify and a repeat buyer will recognize as being “theirs”. Defining flavor is not easy, however.
Flavor attributes are defined qualitatively both in terms that are common to all coffees (such as acidity, body, sweetness, and so forth) and also in terms that are unique to the particular product (earthiness, special origin flavors, particular qualities of acidity). The descriptions used for sensory specifications should relate directly to a particular flavor quality and a “gold standard” that particularly illustrates the quality should be presented to the tasting panel as often as possible as a reference. In many foods, particular flavor qualities are related to the presence of known chemicals. Chemical analysis of coffee is difficult since the flavor components are present in such minute amounts, but some training aids using specific flavor compounds are possible and can be used as references.
Once a list of the necessary flavor attributes is produced, the relative intensities of those attributes should be quantified. A common method within the food industry of expressing the combination of qualities and intensities, especially when developing the standards for a product, is a spider gram (also called a radar gram).
In a spider chart, the various flavor attributes are represented as a circle. The described quality is listed on the outside and the intensity of the particular attribute is seen as the distance from the center. Starting at the top and proceeding clockwise, the order in which the attributes follow one another are elucidated. An example is shown in Figure 1.
In this product, one can see that the emphasis is on earthy flavor, roasted coffee flavor, and heavy body, with acidity playing less of a role. The coffee is not intended to have a long finish, but should have some definite sweetness and caramel qualities.
For this roaster, the flavor profile illustrated in Figure 1 constitutes the ideal Sumatra Mandheling that they would always like to present to customers. However, in the real world, this cannot always be achieved, so ranges for the sensory aspects of the product are also defined. This is shown in Figure 2.
In Figure 2, the flavor attributes of intended product are clarified as the range of attributes are delineated. Some attributes have rather narrow ranges, such as those relating to earthiness, indicating that a certain level of this particular attribute must be present for this product, but too much is not allowable. In other cases, such as those attributes relating to sweetness and body, the minimum amount is close to the standard, but the maximum allowable amount is relatively high (the coffee is unlikely to have too much sweetness or body). With some attributes, the minimum is low and the maximum is closer to the ideal standard, like acidity (this product should not have excessive acidity) and coffee/sulfur attributes. Some of these attributes have negative connotations to consumers (how many would buy a coffee described as “sulfurous” or “bitter”?) but may make important contributions to the overall flavor profile.
Once the flavor attributes of a coffee product are defined, green coffee buyers have a definite idea of what they are looking for. Forms specific to this product can be made up for their use, as seen in Figures 3 and 4. At this point, it is assumed that any defect, including physical defects, would disqualify a sample from further consideration.
Use of either Figures 3 or 4 depends on the preference and style of the cupping panel. Some cuppers prefer a definite numeric system and have psychologically calibrated themselves to this, while others prefer to express intensity using the open linear scale. In the latter case, an overlay is used so that the relative intensity can be recorded (Figure 5).
Note that the green coffee buyer’s form does not contain references to attributes that will occur as the result of roast level. The emphasis is on the flavor attributes that will be found in the green coffee itself.
The type of form to use depends upon the preference of the cuppers and the stage at which the coffee is to be evaluated. The form in Figure 3 (or even a simplified form) could be used in an initial evaluation where numerous potential purchases are being evaluated. It also has the advantage that several can be placed on a single page. If closer determination is needed, the form in Figure 4, which allows the cupper a wider space for finer discrimination between relative intensity, can be used. Another consideration is the amount of time it takes to enter the data produced. As hand-held electronic data entry devices that can be directly downloaded into a computer become more available, this will be less of an issue.
The samples cupped by the green coffee buyers are then compared in terms of the intended flavor attributes of the coffee, as in Figure 5.
As is usually the case, none of the products perfectly conform to the product specification. Sample # 3 is clearly lacking in some flavor attributes. Sample #1 was apparently promising in terms of its caramel aroma and fragrance, but disappointed the panel in terms of its sweetness and body. The sample that conforms most closely to the intended flavor profile is sample #2, which has the requisite sweetness, body, and earthy flavor while not significantly lacking in other attributes. Of course, price of the offers will play a part in the final decision and it may be decided that one of the others is adequate.
Once the product goes into production, the original product specifications are used in evaluating the coffee as it will be presented to the customer. A form for this final evaluation can be seen in Figure 6. Such a form can be used not only in the cupping room but as a “quick check” for roasting technicians to evaluate the batches they are producing.
In this form, some attributes are combined and those qualities that are the result of level of roast are emphasized. The levels of intensity are nominal (given names) rather than numerical, which will be clearer to those who do not cup every day (such as sales, marketing, and some management personnel; such forms will allow their regular participation). The goal of the quality control form is to ensure that all necessary qualities are present while allowing an efficient evaluation.
Summary
One of the most important aspects of a coffee roaster’s quality control program is the sensory analysis procedure known as cupping. For results to be relatively objective and useful, the program must be carefully managed.
Strict testing protocols should be established and followed.
Specifications for products must be made that are specific enough that the flavorattributes that should be in the products are present, but not so detailed that testingbecomes overly difficult or that products cannot measure up to the standards.
Putting these procedures into place is most likely to result in meaningful data that can genuinely reflect the quality of product intended by the roaster.