The decision matrix
A score tells you where something landed. A matrix tells you what moved it.
Every number on this page is already somewhere else in the application — these are the weights the Edit applied to your profile, unchanged, with the row totals still adding up to the scores it showed you. What is new is the shape. Laid side by side, most criteria turn out to move every candidate together, which changes the numbers and decides nothing. Usually one or two are doing all the separating, and it is worth knowing which.
Reading your profile from this browser…
Compare within
Every product type, ranked against each other.
Two is a duel. Eight is a table you will read once.
Figure · the comparison, drawn
- Weight for this candidate— drawn right of the midline
- Weight against— drawn left of the midline
- Decisive column— remove it and the leader changes
What actually decided this
- Performance sport tint and Pressed powder compact are 1 point apart. At that distance the ranking is a coin toss with extra steps — pick on the job you actually need done, not on the order.
- The column doing the separating is coverage match — 5 points between the best and worst candidate on that criterion alone.
- Sebum behaviour moves every candidate identically, so it changes the numbers and decides nothing.
- The lead rests on one criterion. Take coverage match away and a different candidate wins.
Coverage match is deciding this comparison, and you have not set the field behind it — the desk is using its own default. Set it on the profile panel and the ranking here becomes yours rather than the desk's.
Every cell, with its arithmetic
The table is the source; the figure is the summary.
The figure drops criteria that move every candidate identically, because a picture of a constant is a picture of nothing. This table keeps all of them, and prints the weight the engine used wherever it published one.
| Criterion | Performance sport tint | Pressed powder compact | Long-wear cheek stain | Powder blush | Powder bronzer | Spread |
|---|---|---|---|---|---|---|
| Neutral startbaseline | 50 | 50 | 50 | 50 | 50 | 0 |
| Coverage matchdecisive−|Δ coverage| × 0.42 | -3 | -4 | -8 | -6 | -6 | 5 |
| Sebum behaviourdecides nothingoil × 3.5 | +7 | +7 | +7 | +7 | +7 | 0 |
| Film cost−layerWeight × 4 | -4 | -4 | — | -4 | -4 | 4 |
| Score | 50 | 49 | 49 | 47 | 47 |
A wide spread means this desk’s model discriminates strongly on that axis. It does not mean reality does. These are weights on a preference, not measurements of a face, and no cell here is a probability.
core
Performance sport tint
Thin pigment that survives sweat without inviting powder.
Open the full note →consider
Pressed powder compact
Portable set — the classic afternoon cake mechanism.
Open the full note →consider
Long-wear cheek stain
Colour that survives humidity and sweat without a cream film.
Open the full note →Watch the hand · 4 referencesWhat the winning architecture looks like on a face.
Chosen to demonstrate a decision this desk already makes. They are technique references, not endorsements, and a product named inside one is an archive example rather than a recommendation.
Where Should Blush ACTUALLY Go?
Robert Welsh
Tests high, low and forward placement on one face, so placement alone visibly changes face length.
Why this source · Direct side-by-side comparison; no product change between passes.
HOW TO APPLY HIGHLIGHTER TO LOOK BEAUTIFUL - NOT FAKE!
Wayne Goss
The bone-top placements light actually hits, and the adjacent spots that read as grease.
Why this source · Long-running technique channel; placement demonstrated against the face, not a diagram.
What is foundation for?
Alexandra Anele
Reframes foundation as an evening layer rather than a mask, so you can see how little actually does the job.
Why this source · The coverage-appetite question, argued from the object's purpose.
Bronzer vs Contour - Seamless application for everyday
Lisa Eldridge
Runs bronzer and contour on the same face so structure and warmth read as two placements, not two products.
Why this source · Thirty-year professional makeup artist; the desk's roles module makes the same distinction.
Next
Where to take this
Chosen from your own record: 3 things open at the beginner stage, and what usually follows this page.
Back to the Edit
Change one answer and watch which column stops deciding.
Two, side by side
When the matrix says the top two are a coin toss, this is where the jobs get compared rather than the scores.
Log one worn day
Log a day
0 of 1 so far. What held past hour six is the one fact no formula list can predict for you.
When this is not a makeup question
Nothing is written into another desk without you choosing it. These links carry only the context named on them.
- Skincare Intelligence
Carry your skin, climate, sensitivity and named concerns across so the routine can be evaluated with the same stated inputs the makeup desk used.
- Vanity or Vice · Routines & Ingredients
The order you apply things in, what each film is intended to do, and where the evidence is uncertain — argued long-form rather than scored. This is the reading behind the sequence this desk just gave you.