Concentric dry-brush contour rings radiating from a single point, in the manner of a smellmap.

Smellmaps for perfume

Research 5 min read

We took the visual grammar Dr Kate McLean developed for mapping the smell of cities and applied it to every perfume we hold. This page explains what we borrowed, exactly which of our data drives it, where the analogy stops - and what we added.

On this page
  1. The research
  2. What we mapped it onto
  3. Where the analogy stops
  4. What we added
  5. See it
  6. Keep going

The research

Dr Kate McLean is an artist, designer and researcher, an honorary research fellow at the University of Kent, and is writing her first book, an Atlas of Scents, Smells and Stinks, for Laurence King, expected in 2027. Over the past fifteen years she has produced smellmaps of cities including Paris, Amsterdam, New York, Singapore, Milan, Marseille and Kyiv. Each one is built from participatory smellwalks, in which local people log the odours they meet and rate each one for intensity, pleasantness and familiarity.

What makes the work matter to us is the encoding. A smellmap is not decoration. Hue carries which smell it is. The size and density of a bloom carry intensity and dominance. Diffusion across the map carries how the smell travels on the air, and a base-versus-episodic distinction carries persistence. It is a visual language for a thing you cannot see, and it is honest about being an impression rather than a measurement - which is exactly the discipline a perfume visualisation needs.

Four variables adapted from McLean's work, in ScentVerdict's rendering and in the colours our maps really use. The table below says which of our stored numbers drives each one.
Dior Sauvage, drawn from the same profile and by the same code as its own live smellmap. Everything the key above describes, assembled.

What we mapped it onto

Every perfume in our catalogue carries a structured profile: its notes, each with a phase and a weight, a colour and a scent family per note, and three performance scores rated one to five. The map reads most but not all of that: notes, phases, weights, colours, and two of the three scores. Intensity scales the marks and projection scales the rings. Longevity it never looks at. Where a value is missing it falls back - a note with no phase is treated as a heart note, a note with no colour takes its family's, and a missing score is read as three out of five.

McLean's encoding Our data On the smellmap
Hue - which smell it is Note colour, falling back to its scent family The colour of each bloom
Size and density - intensity and dominance Note weight within the perfume, scaled by the perfume's intensity score Mark radius and opacity - and the stronger a note, the higher it sits within its lane
Diffusion - how far it travels Projection score, one to five The number of contour rings and how far they reach. Reach is also scaled by the mark itself, so a big note carries further than a small one at the same projection
Base versus episodic - persistence Note phase: base, heart or top Three lanes across the canvas, top at the left through to base at the right. Top notes are crisp round marks, base notes pool into soft flat washes. Within a lane the horizontal position is random, so it carries nothing
Impressionistic register - honesty about precision A stylistic choice, not a data field Soft edges and translucency throughout - the map never claims to be a measurement

Scroll the table sideways to see every column

Where the analogy stops

McLean's maps are hand-crafted artworks built from crowd-sourced human perception on a specific day in a specific place. Ours are generated deterministically from a structured profile, the same every time, across thousands of perfumes. We have borrowed her grammar, not her method, and we make no claim to her craft. The colour, size, form, ring reach, phase lane and height within the lane all carry data. The position within a lane does not, and neither does the small jitter that stops marks landing on a grid.

The soft register is deliberate and load-bearing. A separate strand of research, Kateřina Milatová's 2025 thesis on visualising fragrance online, found that a vivid scent visualisation raises a buyer's confidence without reliably raising their accuracy. An obviously impressionistic map makes a promise it can keep. See our page on that work.

What we added

Three departures are ours, made deliberately rather than by drift.

Time in place of space. A smellwalk traverses a city. A perfume traverses hours on skin. We kept McLean's grammar and swapped the dimension it maps. Our canvas reads left to right, first spray to drydown - a smellwalk through a perfume rather than through a place. Milatová's usability work independently points the same way: she replaced the vertical note pyramid with a horizontal evolution timeline, and her participants rated it the second most helpful cue on the page.

Named marks. McLean's participants walk the streets her maps describe. A perfume buyer has no such ground truth. Milatová found that colour alone is not enough to make a scent legible online: her participants preferred images to colours, and not one of them reached for abstraction when asked to draw a scent. So every note is named directly on the map, sized by its weight in the composition. A separate study of AI-made scent images reached the same conclusion from the other direction: an abstract picture with no words attached was the worst of every option they tested. See our page on that work.

One grammar, every perfume. Because the encoding is fixed and the rendering deterministic, the same perfume redraws identically and every map in the catalogue uses the same key. That makes two maps comparable as impressions. It does not make any single mark a scale you can read off: a bloom's size combines the note's weight with the perfume's intensity, and its rings inherit that size as well as the projection score. It is the difference between borrowing her paintings and learning her grammar.

What the map cannot tell you

It has never met the perfume. There is no formula behind it, no measured evaporation curve, and nothing about how it behaves on your skin rather than anyone else's. It draws our stored profile through three broad phase lanes and a fixed seed, so it shows you the shape of what we hold, not a record of what happens hour by hour. That is also why every mark stays soft-edged.

See it

Six perfumes spanning the families, each rendered live from its own data.

Every perfume on ScentVerdict has one - append /smellmap to any perfume URL.

Keep going