Formula × Brush Engineering #003 · A brand doesn’t choose a number of brushes. Its formula lineup already chose.
The question usually arrives expecting a number: should we launch with five shapes? Eight? Twelve? As if there were a correct figure somewhere, waiting in a template.
From the development bench, the honest answer is that the number isn’t chosen at all — it’s derived. The formula defines the job before the brush is drawn; a lineup of formulas defines a map of jobs. Walk the derivation and the number falls out on its own — different for every brand, and correct for each of them.
Step 1 — list the formulas. Everything you sell and everything in the development pipeline: the liquid foundation, the skin tint, the cream blush, the pressed powder, the brow product. The brush count starts here, not in a brush catalog.
Step 2 — name each formula’s job: its behavior, and where it works. Not its marketing name — its mechanics: powder wants pickup, cream wants control, liquid wants even release, balm wants flexibility, sharpened by coverage and finish — and the zone it’s applied to. The same texture swept across a cheek and worked into the corner of an eye is two jobs, not one: same behavior, different precision.
Step 3 — merge and split by the same test. Two formulas that genuinely behave the same way, in the same zone, are one job — and a single behavior serving two zones a customer treats differently is two. The ruler for both directions is the one that governs everything in a line: would the customer feel the difference in use? Felt differences split jobs; unfelt ones merge them — which is exactly why minor size variants don’t count as separate shapes.
Step 4 — one job, one shape. Each distinct job that survives Step 3 earns exactly one engineered shape. That’s the whole answer. A complexion-focused skin-tint brand might derive three shapes and be complete; a full-face artistry line might derive ten and carry no excess. Neither number came from a template. Both came from the catalog.
The map guards both borders at once.
Against too many: any shape that can’t point to a job on the map is a future duplicate — the Functional, Perceptual, or Cosmetic Overlap that accumulates when shapes are added faster than jobs. That article is this one’s mirror: it diagnoses a line that already drifted; the derivation is how you build one that doesn’t.
Against too few: the opposite failure is the “universal” brush stretched across jobs it wasn’t built for — a single head asked to serve a liquid foundation and a pressed powder equally well, which usually means serving neither. That gap doesn’t show at launch; it shows in the first-use complaint window, as streaks and eaten product, billed to a formula that was never the problem.
The derived number is simply the point where neither risk is built into the line itself. Execution still has to be good — a well-derived shape can still be badly built — but the architecture stops working against you.
Plenty of strong brush brands don’t. The derivation holds — only Step 1 changes: instead of your own catalog, list the routine you’re serving — the products your customer already owns and reaches for daily. What does she apply in the morning, in what textures, across which zones? A tools-first brand isn’t exempt from the map. It’s reading someone else’s shelf instead of its own.
The map updates on one trigger only: the lineup of jobs changes. A new product with a genuinely new application behavior — or a new zone — earns a new shape; that’s a brush replacing a compromise, not joining a crowd. A new colorway, a seasonal collection, a trend silhouette — none of those add jobs, so none of them add shapes. The trend calendar doesn’t edit the map. The catalog does.
Two decisions sit downstream of the map, and they’re separate articles for a reason: which of the derived shapes to launch first is a sequencing question, and whether they sell as singles or a set is a format question. Derive first; sequence and package after. Run it in the other order and you’re choosing formats for shapes that may not deserve to exist.
How many brush shapes does a brand really need? As many as its lineup has distinct jobs — no more, no fewer. The number isn’t a strategy. It’s a property of your catalog, waiting to be read.
If you’re planning a line now, don’t send us a brush count — send us the formula list: what each product is, how it behaves, where it’s applied, who applies it. We’ll derive the map with you, shape by shape. Sometimes it comes out smaller than you planned — we’ll say so. Sometimes a job is hiding in the lineup that deserves a brush nobody sketched yet. We’ll say that too.
As many as its lineup has distinct jobs – no more, no fewer. The number is derived, not chosen: list every formula, name each one’s job as its application behavior plus the zone it’s applied to, merge jobs the customer can’t tell apart in use, and give each surviving job one engineered shape. A skin-tint brand might derive three shapes and be complete; a full-face artistry line might derive ten and carry no excess.
Yes – when the products genuinely share a behavior in the same zone. Two liquids with the same coverage, finish, and working area are one job, and forcing two brushes onto them creates overlap. But the same texture applied across a cheek and into the corner of an eye is two jobs despite identical chemistry: same behavior, different precision. The ruler is whether the customer would feel the difference in use.
No. Formulas that behave the same way in the same zone merge into one job, so a ten-product lineup can derive to five or six shapes. The count compresses wherever behaviors genuinely overlap – and only expands when a formula introduces a behavior or a zone the existing shapes weren’t built for.
On one trigger only: the lineup of jobs changes – a new product with a genuinely new application behavior, or a new zone the customer treats differently. A new colorway, a seasonal collection, or a trend silhouette adds no jobs and therefore no shapes. The trend calendar doesn’t edit the map; the catalog does.
The universal-brush failure: one head stretched across jobs it wasn’t engineered for – asked to serve a liquid foundation and a pressed powder equally, and serving neither. It looks fine at launch and surfaces in the first weeks of real use, as streaking and eaten product, in reviews billed to a formula that was never the problem.
