The Methodology for Measuring Liking Scores in Headspace

Liking scores measure how much a consumer is likely to enjoy a product based on its taste and aroma. Using digital twins trained on extensive datasets, Headspace predicts liking with precision, helping brands craft products that resonate with their target audience.

How Liking Scores Are Determined

Liking scores are given by Headspace’s digital twins, which are virtual replicas of target consumers. These digital twins are trained to virtually taste blindly a product, providing a score that reflects real consumer panel results. The process integrates:

• Scientific Data: Products are analyzed in Headspace’s lab to identify and quantify aroma molecules, which are then linked to consumer preferences using the knowledge graph.
• Tailored Models: Liking scores are region- and audience-specific. For example, a digital twin for health-conscious Gen Z consumers in the U.S. differs from one representing the general population in France.

How Foodpairing Validates the accuracy Liking Score

Foodpairing validates its Liking scores by comparing predictions from digital twins to results from blind consumer panels and surveys. Across multiple tests, the models consistently show 80%–85% correlation, often reaching up to 97%, with a typical error margin under 5%.

Backed by a sensory team and continuous benchmarking, Foodpairing ensures its AI predictions are scientifically grounded and reliably aligned with real consumer preferences.

Key Features of Liking Analysis

1. Scientific Precision
   Liking scores are grounded in the chemical analysis of products, ensuring scientifically validated predictions.

2. Regional and Audience Customization
   Recognizing that preferences vary across regions, Headspace builds digital twins tailored to specific markets and demographics.

3. Identifying Drivers of Liking and Disliking
   Beyond the score, Headspace highlights aroma attributes that drive liking or disliking. For instance, a product might benefit from more lemon aroma while reducing cinnamon.

4. Validation Through Consumer Panels
   Predicted scores are compared against actual consumer panel results, with correlations between 80% and 95% and an error margin below 6%. This ensures the accuracy and reliability of the predictions.

Why Liking Scores Matter

Liking scores are essential for successful product development, providing:

• Optimization: Brands can refine formulations to align with consumer preferences.
• Consumer Appeal: Insights into what drives liking help create products consumers truly enjoy.
• Informed Decisions: Combining scientific and consumer data ensures data-driven innovation.

Conclusion

Liking scores are a vital tool for understanding consumer preferences and creating products that delight. By combining advanced data analysis, tailored digital twins, and rigorous validation, Headspace empowers brands to innovate with confidence, ensuring their products not only meet but exceed consumer expectations.