Why it smells flat when every material is good
Flat is not a materials problem. It is too much agreement. The bench puts a number on it, and the number has a band to sit in.
You can build a dull perfume out of fifteen good things. It happens the same way every time: every line does roughly the same job, so there is nothing for the nose to travel between, and a perfume that does not move reads as one note.
Flat is not the same as small. A small perfume is quiet and you want more of it. A flat one can be loud, last twelve hours and still bore you, because hour six smells like minute one.
The number on the bench
The Reading panel puts a figure on this. Cohesion is the average facet overlap between every pair of materials in the formula, weighted by how loud each one is, on a scale of 100.
| Cohesion | Reads as | What it means |
|---|---|---|
| Under 6 | Loose | The materials barely know each other |
| 6 to 9 | Average | Held together loosely. Chypres, fougères and ouds live here by design |
| 9 to 13 | Coherent | The band most finished formulas sit in |
| 13 to 24 | Very tight | Everything agreeing with everything. Typical of woody ambers |
| Above 24 | Flat | So much shared ground that the perfume has nowhere to go |
Balance cohesion aims at 11 and pushes in either direction to get there, adding a connector when the formula is loose and contrast when it is tight.
How a pair earns its score
Two materials score by the facets they share divided by all the facets between them, with a bonus for the same family. Worth seeing once with real numbers.
Iso E Super is filed as woody, velvety, ambery and transparent. Georgywood is woody, ambery, warm and velvety. Three facets shared out of the five between them, both in the woody family: that pair reads 72. Iso E Super against Timbersilk reads 52. Oakmoss against any of the three reads zero, because mossy, earthy, phenolic and inky have nothing in common with a transparent wood.
Against a target average of 11, one pair at 72 is not a detail. In a formula of eight equal parts that single pair puts about two and a half points on the average by itself, and more if those two are the loudest things in it.
The connectors are usually the cause
Iso E Super has a working window of 2 to 40 per cent of the oil. Timbersilk 2 to 35. Hedione 2 to 40. The windows are that wide because none of the three is meant to state anything of its own: they fill the space between the notes you already have. The library tags them as bridges, along with Ambroxan, Linalool and Habanolide.
A material that shares ground with everything raises the average by definition. Put two or three bridges in at double figures and you have not added notes, you have added agreement. The Iso E Super post covers what that material does for volume, which is real: space between materials and shared ground between them are not the same measurement.
Read it against what you are building
A woody amber is one family end to end and will sit well up in the Very tight band with nothing wrong with it. An oud sits low because the contrast is the point. If the figure is high and the perfume smells like the thing you meant, leave it.
Three moves when it is too high
- Halve the loudest bridge. Halve it, do not remove it: the volume comes from that material, and with the woody bridges much of the longevity too. Read the number, then smell both versions the next morning.
- Bring in something that scores zero. One deliberate outlier does more than three small adjustments. Oakmoss absolute low-atranol runs 0.05 to 1.5 per cent of the oil, capped by IFRA at 0.1 per cent of the finished product in category 4, which at eau de parfum strength bites well before the top of that window. cis-3-Hexenol, which is cut grass, runs 0.005 to 0.4 per cent with no cap at all. Neither shares a facet with a transparent wood.
- Turn up the outlier you already have. A tight formula usually has one material being drowned by the rest. This is the move the bench makes itself when the figure is too high: it finds the material that agrees least with everything else and multiplies it by 1.8, then clamps the formula back inside its dose windows.
Where the number stops
Cohesion says nothing about whether the formula is legal or whether anything is overdosed. Those are separate readings, and a perfume can be coherent and over three limits at once. It says nothing about shape either: a formula can sit at 11 with only 8 per cent of its perceived weight in the top, which is its own kind of flat, and one the pyramid reading catches instead.
Put a formula you already like on the bench, note the cohesion figure, then press Balance cohesion and read the line it writes back. It names what it brought in or turned down and why, in words rather than numbers. Undo puts it back if you disagree with it.
Open the bench →One honest note about the band
The 9 to 13 band was measured, not decided. It is where the middle half of the formulas the studio generates happens to fall, which makes it a description of what usually works rather than a law.