Challenge

Traditional sensory market research typically involves consumers physically testing products and providing feedback, often through Central Location Tests (CLTs) or Home Use Tests (HUTs). However, consumer perception begins well before the tasting—it is shaped by expectations formed through product descriptions, past experiences, and marketing cues such as branding

Approach

Introducing a novel methodology to assess consumer preferences by capturing projected product liking based solely on a comprehensive sensory-focused description.

Step 1 – Blind Product Testing (Control Group)

Blind Central Location Test with n=168 consumers of the category, 4 yoghurts evaluated.

Outputs: liking scores

Step 2 – Sensory Profiling

QDA expert panel to measure the sensory characteristics of the 4 yogurts.

Outputs: sensory profiles & differences between products. Statistical crossing with Step 1 to give defined drivers of liking

Step 3 – AI-Language Generation

Sensory profiles from Step 2 were used as input for a Large Language Model (LLM)-based AI-tool to produce 2 types of consumer-oriented descriptions:

• a marketing-style version (emotionally engaging / brand-like language)
• and a formal-style version (neutral / descriptive language).

Outputs: 2 descriptions (marketing + formal) for each of the 4 yoghurts.

Step 4 – Online Consumer Evaluation of Projected Liking

n=438  yogurt consumers read the 2 AI-generated descriptions of all 4 products in a monadic sequence.

Outputs: anticipated liking scores; projected sensory experience; feedback on descriptions.

Outcome

Marketing-style descriptions received slightly higher ratings, suggesting a preference for persuasive language. Product rankings from descriptions closely matched tasting results.

Sensory cues influenced expectations but not actual liking. Conversely, “fat” was a key driver during tasting but undervalued in descriptions, highlighting a disconnect between language and sensory perception. 

The “Read & Rate” method offers useful insights into how narratives shape expectations, but its predictive accuracy is limited. Future research should refine the approach, expand product categories, and explore cultural differences in interpreting descriptive language.

Click here to view our Pangborn 2025 research poster.