Beyond Simple Attribute Ratings
Developing a more informative approach to sensory and consumer product characterization
Simple attribute-based methods can make product evaluation faster and more accessible. But when products differ only subtly, simplicity may come at the cost of discrimination, response quality and interpretability.
The Double-Faced Applicability test was developed to retain the practical advantages of structured attribute evaluation while generating richer information about how products are perceived and experienced.
The Challenge
Methods such as Check-All-That-Apply allow consumers to describe products without extensive training. However, an unchecked attribute can be difficult to interpret. It may mean that the descriptor does not apply, was overlooked, or was not considered carefully.
Rating-based variants provide more information, but can introduce variability in scale use. Conventional descriptive analysis offers detailed characterization, but requires product-specific training and substantial investment in panel development.
The methodological question was therefore:
Could an attribute-based task remain accessible while supporting more careful evaluation, stronger product discrimination and more interpretable profiles?
The Approach
The Double-Faced Applicability test, or DFA, uses a two-step response.
For each descriptor, participants first decide whether it applies to the product:
Yes — it applies
or
No — it does not apply
They then indicate how certain they are about that decision.
DFA also evaluates contrasting descriptors separately — for example, convenient to use and inconvenient to use — rather than assuming they are exact opposites.
Read more
This response structure combines explicit applicability judgements with graded certainty. Positive and negative descriptors are evaluated independently, allowing the method to capture situations where one side does not automatically imply the other. The approach was developed using principles from Signal Detection Theory to support more stable and interpretable attribute evaluation.
From Product Differences to Product Profiles
DFA was designed not only to improve the response task, but also to support a different way of profiling products.
By combining responses to the positive and negative sides of an attribute, each product can be characterized independently.
This makes it possible to see:
which side best describes the product
how strongly it applies
how products compare
how the experience changes across usage stages
The result is more than a ranking between products. It helps explain what each product is actually like.
Read more
To support this interpretation, the DFA responses were summarized using an SDT-based applicability measure, d’A. This provides an independent estimate for each product and attribute, rather than only expressing differences relative to a reference product.
Positive values indicate that the positive descriptor is more applicable, negative values indicate that the negative descriptor is more applicable, and values around zero indicate that neither side clearly dominates at group level.
The measure can also be calculated for different usage stages, making it possible to build profiles that show how the product experience changes over time.
Application in Consumer Research
DFA has been applied in both non-food and food consumer research.
In household cleaning, it was used to characterize how consumers experienced products across different stages of use.
In the mayonnaise study, DFA was used both before tasting, to capture consumer expectations, and after tasting, to characterize the sensory experience of six products.
Together, these applications show how the same underlying approach can be adapted to different product categories and different stages of the consumer experience.
Read more
In the household cleaning work, DFA was used to follow the consumer experience through different stages of product use and identify where product differences emerged.
In the mayonnaise research, DFA was used more broadly as part of an expectation–experience framework. Consumers first evaluated what they expected from mayonnaise in different consumption situations, and later evaluated the sensory characteristics of actual products. The expectation profiles were expressed quantitatively using d’A, allowing consumer expectations and experienced product characteristics to be compared on the same conceptual framework.
A later analysis of the mayonnaise data showed that the DFA-based expectation profiles identified attributes associated with overall satisfaction and broadly aligned with results from conventional approaches such as PLS regression and Landscape Segmentation Analysis. The expectation profile also extended beyond the existing product set, helping identify directions for product optimization and opportunities for innovation.
Application in Sensory Profiling
DFA was also evaluated as a faster approach to sensory profiling.
Twelve commercial spread products were assessed using both conventional descriptive analysis and DFA. Despite a major difference in category-specific preparation, the two approaches produced highly similar product configurations, with an RV coefficient of 0.93.
Conventional descriptive analysis remained somewhat more sensitive in pairwise product discrimination, but DFA reproduced much of the same sensory product structure with substantially less training.
This suggests that DFA can offer a more efficient route to sensory characterization when an appropriate attribute list is already available.
Read more
The descriptive panel had 10–25 years of sensory experience, including extensive experience with spread products, and received a further 7.5 hours of refresher training in attribute use, scaling, discussion and monitoring before testing. The DFA panel was also experienced in sensory evaluation, but had no spread-specific experience or training and received only a 10-minute introduction and a warm-up sample.
Across individual attributes, the average correlation between the two approaches was 0.72. Conventional descriptive analysis showed somewhat greater pairwise discrimination, while the overall product configurations were highly similar.
The main efficiency gain demonstrated in this study therefore lies in reducing category-specific preparation and training, rather than necessarily reducing evaluation time itself.
What the Research Demonstrated
Across the studies, DFA provided:
More explicit responses
Participants indicate whether an attribute applies and how certain they are.
Greater response differentiation
Certainty ratings added graded information without relying on a conventional intensity scale.
Independent evaluation of contrasting descriptors
One side does not have to imply the other.
Independent product characterization
Using the d’A measure, products can be profiled without requiring a physical reference.
Broad applicability
Demonstrated with consumers and sensory panels, across food and non-food products.
The Decision-Making Value
The value of a product-characterization method lies not only in how quickly data can be collected, but in whether the evidence is sufficiently informative for the decision.
DFA offers a middle ground between simple consumer attribute tasks and fully trained descriptive analysis.
Depending on the research objective, it can support consumer research, prototype comparison, product optimization and efficient sensory profiling.
Key Learning
The way an attribute question is asked determines the information that can be extracted from the answer.
By separating applicability, certainty and contrasting descriptors, DFA provides more information about how people characterize a product without automatically requiring a more elaborate research design.
The broader lesson is simple: choose the measurement approach according to the information the decision requires.
