The research behind ScentVerdict
Showing what a perfume smells like is a real research problem, and people have studied it properly. These are the papers we built on. Each write-up says what we took, which of our data drives it, where we disagree, and what the research caught us getting wrong.
Making a smell visible
- Why we show a picture of the raw ingredient A picture of a scented thing makes you half-smell it, and that changes what people buy. Sharma and Estes, 2024.
- Why every note has a colour Ninety years of colour-smell research, used to audit our own palette, including the two notes we disagree on. Spence, 2020.
- Can AI paint a smell? A Tsinghua team turned 20 real smells into 560 pictures, then asked 28 people how well each one matched. Yu and others, 2026.
- Smellmaps for perfume The visual grammar built for mapping the smell of cities, applied to every perfume we hold. Perkins and McLean, 2020.
- Visualising fragrance online A measured study of what helps someone imagine a perfume they cannot smell, and what it warned us about. Milatová, 2025.
Measuring and describing smell
- Can you measure whether two perfumes smell the same? Two labs built a way to score how alike two smells are, one from molecules and one from a neural network. We do it a third, much cruder way. Snitz and others, 2013; Lee and others, 2023.
- Can you actually describe a smell in words? English speakers are bad at naming smells, and that turns out to be a fact about English, not about people. Majid, 2021.