For every hair type, 1a to 4c
CurlCue
A hair-care app that scores a product against your hair.
Scan a bottle and get a number built from your porosity, your scalp and the look you are steering toward — not from one universal good-or-bad list.
Because some products work against your hair, and nothing on the shelf tells you which.
iOS · in development · the phone is live, tap the mark
Scan
SCORING FOR
Wavy, curly or coily (2–4) · low porosity · balanced scalp
Same bottles, different hair. Every number is the app’s own engine run against that profile — porosity and scalp are what move it, and a health gate can veto a score outright.
Three pillars, not one scanner
Scanning is how it starts. It isn’t what it is.
Point it at a bottle. Get an answer about your hair.
Barcode for anything in the catalogue, or photograph the ingredients list and it reads them. Scoring runs entirely on the device, so the answer arrives in the aisle rather than after a round trip.
Miss a product and the miss becomes the ask — photograph it once and it is there for the next person.
How the score is made
A health floor, then a direction.
Health is the floor. The look is the direction you steer once you are standing on it.
The health gate
Does this help or harm your hair and scalp — buildup, stripping, irritation, real hydration? A bad gate vetoes the score no matter how well the product serves the look you want. A dry scalp weights irritation differently from a balanced one.
The look alignment
Then, how far does it move you toward what you actually want? Eight axes — volume, shine, hold, frizz, cast, moisture, scalp, repair — weighed differently depending on what the product is for. A shampoo cannot add volume, but it can flatten it.
Three lookups per ingredient
What is it doing in this product, does it belong in this category, and how good a version of its job is it? Caprylyl glycol and methylisothiazolinone are both preservatives. They are not the same news.
Most scanners give every bottle one universal score. That is the thing this engine is built not to be.
The score scale
Drag it. A state answers “what happened”; a scale answers “how much”. Different questions, so different colours — and the climb never doubles back.
What CurlCue won’t do
- Tell you a product is “clean”.
- Rate everything 80.
- Pretend one score fits every head of hair.
Two bottles, six people
The cheaper one wins. Until it doesn’t.
Two leave-ins, same job — one of them ten times the price. Scored against real profiles by the app’s own engine, and they change places.
Cantu
Comeback Curl Next Day Curl Revitalizer Spray
Leave-In · $0.50/oz
Olaplex
No.5 Leave-In Moisturize & Mend Leave-In Conditioner
Leave-In · $5.15/oz
low porosity
balanced scalp
72
Cantu
40
Olaplex
CAPPED · −50
CANTU WINS
medium porosity
balanced scalp
67
Cantu
91
Olaplex
OLAPLEX WINS
high porosity
balanced scalp
72
Cantu
93
Olaplex
OLAPLEX WINS
WHY IT FLIPS
The health gate. On low-porosity hair the Olaplex trips protein overload, and 50 points come off a formula that would otherwise score 90 — the best number on this page. It scores 40, and the spray costing a tenth as much beats it by 32. Raise the porosity and the same bottle passes clean and wins by 24, then by 21. Nothing about the bottles changed. The cuticle did.
What does your hair do on its own?
When a water drop hits a dry strand, it…
And how do you want it to sit?
Answering scores this whole page for you. In the app there is a wash-day version of the same question for anyone who would rather answer that one — it exists because an onboarding audit ran this version past four test profiles and two could not answer it.
What the money buys
Almost none of it.
Two bottles changing places could be luck. So the same question went to the whole catalogue.
Every product in the catalogue with both a real price per ounce and a parsed ingredient list — 839 of them — scored against one person, then sorted by what they cost.
Price explains 1.1% of how well a product fits her. The median climbs five points between the cheapest shelf and a dollar an ounce, and then stops. Above eight dollars it gives the five points back.
The catalogue’s cheap end has bad rows in it — a $1 Redken conditioner, an Olaplex bond builder priced as a 69-ounce bottle — and cheap junk drags a correlation toward zero, which is the direction this page would prefer. So it was run three ways: everything (1.5%), without the inconsistent rows (1.1%), and without those and everything under $3 (1.1%). It lands in the same place whichever way, which is the only reason it is here.
SOLID = MEDIAN SCORE IN THE BAND · PALE = BEST IN THE BAND
The best product under a dollar an ounce scores 85. The best between four and eight dollars scores 91. Six points, for eight times the money — and only if you pick the right one, because the median in that band is 69.
CurlCue asks your budget in onboarding and stores it. The engine does not read it. Nothing in the scoring path touches that field, so a score here is not adjusted for what you are willing to spend — and this section is what happens when you check whether it would need to be.
What happens after the number
CurlCue AI keeps the score in front of it.
Every answer starts with a number it didn’t choose. The engine decides the score. The model is handed that score, the axis breakdown, the specific reasons it moved and your profile — and then has to explain itself against them. It cannot overrule the number, and it is told in as many words never to invent one.
ONE BOTTLE
Coconut & Hibiscus Curl & Shine Conditioner
SheaMoisture · Conditioner
WHOSE HAIR
WHAT THE ENGINE HANDED IT
Strong: Moisture, Scalp Tolerance
Weak: Repair / Strength
4C · low porosity · coarse · dry scalp
WHAT IT WROTE BACK
…watch out for the fact that coconut-heavy formulas can sit on top of low-porosity strands rather than penetrating them, which means your dryness goal may go unmet even after consistent use.
CAPTURED 4 AUGUST 2026 · ENGINE 2.1
The 4C and the 1C both score 65, and the model objects for opposite reasons — on low-porosity coils the formula never gets in, on a wave the same richness weighs it down. One number, two unrelated failures. That is why a score here is a fit, not a grade.
Real rows from the app’s own verdict cache, captured between May and August 2026 on engine versions 2.0 and 2.1. The engine is on 2.7 now, so rescoring today would move these numbers — they are real output, not current output.
ONE REAL EXCHANGE, TRANSCRIBED
WHAT IT IS HANDED
IT CANNOT OVERRULE THIS
PRODUCT
Coconut & Hibiscus Curl & Shine Conditioner
SheaMoisture · Conditioner
Strong: Moisture, Scalp Tolerance
Weak: Repair / Strength
For 4C low-porosity coarse hair, moisture delivery is everything — and watch out for the fact that coconut-heavy formulas can sit on top of low-por…
AND WHO IS ASKING
4C · low porosity · coarse · dry scalp · washes weekly · long
WHAT IT SAYS BACK
Should I use this, swap it, or pair it?
00:44
Honest take: a 65 is a workable score, not a red flag — but the verdict is pointing at something real for your specific hair. 💧
With low-porosity coarse 4C coils, your cuticle is tight and resistant. Coconut-heavy conditioners tend to sit on the surface rather than penetrate, which means you might finish wash day feeling coated but not actually moisturized — and that dryness creeps back fast. For a weekly wash schedule and long hair, that's a real gap to carry for 7–10 days.
AND THEN IT HANDS YOU BACK TO THE APP
THE ARROWED TWO WORK HERE, TOO — THEY JUMP TO THIS PAGE’S OWN COMPARE GRID AND LIVE DEMO, NOT TO A COPY OF THE APP
IT OFFERED THESE THREE. THEY SENT ONE
One real exchange, transcribed from the app on 17 August 2026 — emoji included, and including the fact that its answer told this person to keep the bottle they already own. Nothing here is scripted, and nothing here is generated for the page: every control only unfolds text already in the transcript. The three questions are the ones it offered; the one that opens is the one they actually sent. The score was 65 at capture; the engine is on 2.7 now and rescores this bottle at 64.
Measured, not estimated
1,785
ingredients classified
by role, category fit and quality tier
1,467
products in the catalogue
132 brands, built by the people using it
22,302
product–ingredient links
19.6 parsed ingredients per product
12
hair types, 1a to 4c
not a curls-only app
22
signals collected
every one of them read by the engine
1,139
of those can be scored today
the rest are waiting on an ingredient list
66
engine assertions
plus a scoring ratchet that can only tighten
Counted against the production database on 12 August 2026 and re-verified unchanged on 17 August. Every score on this page is engine 2.7 — except the AI verdicts, which were captured earlier and carry their own version where they appear.
Every screen is a real capture
Taken from the running app, not mocked up.
Captured 17 August 2026 on engine 2.7 — tap any of them to read one properly. Including the one that is empty: the social layer is built and switched on, and nobody is in it yet.
Home
Your streak, your journey, and the mark that starts a scan.
Product
Why your score is what it is, ingredient by ingredient.
Routine
Six steps generated from one profile, each explaining itself.
Community
Built, running, and empty until there are people in it.
Compare
Two products side by side, with the tradeoffs named rather than averaged away.
Not shipped yet
It isn’t out yet.
CurlCue is in development on iOS — no App Store listing, no TestFlight, no waitlist to put you on. If you want to know when it lands, say so and I’ll tell you. That is the whole list.