Skincare that starts with a measurement.
Twenty-seven things get measured — optically, at the surface, in the living film on your skin, in blood, in urine, in how fast you recover from a deliberate small insult, and in what the sun actually did to you. Nothing is guessed from a photo.
Skin, and the scale it gets measured against. The six bands are ITA° exactly as Chardon defined them in 1991 — one axis of twenty-seven. Colour alone has never been enough to build from.
axis 1 / 27 · computed, not selected
The Apparatus · seven instruments
One arrives in a box and goes back in the same box.
A phone camera has an uncontrolled light source, automatic white balance and a sensor that guesses at colour. It cannot measure skin. So we ship something that can — and six more things it can’t do alone. Every method below is published; open any card for the source.
The Reader
ships to you · returns to us
Five sites, ninety seconds. Colour, redness, thickness and water loss — measured optically, not estimated.
The Reader
- ITA°arctan[(L*−50)/b*]·180/π, from reflectance CIELAB
- MMelanin index, narrow-band reflectance
- EErythema index, same optics
- µmEpidermal thickness, optical coherence tomography
A closed-chamber evaporimeter for transepidermal water loss in g/m²/h — the barrier measurement the strips cannot make.
OCT at this size is the engineering claim in this kit. The optics are clinic-grade today and we will keep saying so until the unit on your counter matches the one in the study. TEWL also drifts with room humidity and airflow, so the app holds the reading until conditions sit inside the window.
Chardon, Cretois & Hourseau 1991 — ITA° and the six-band classification. Bands as published: very light >55°, light 41–55, intermediate 28–41, tan 10–28, brown −30–10, dark <−30.
Surface strips
you keep · read by the app
The only four read by camera — and each sits beside a printed reference in the same frame.
Surface strips
- pHSurface pH by litmus, three sites
- AUSebum as optical density on sebum-absorbent tape
- idxDesquamation from an adhesive disc
- idxCorneocyte scaling, same disc
A reference in-frame turns a camera from a guesser into a comparator. It is never measuring absolute colour — it measures a difference against a known patch under the same light, which is the only thing a phone sensor can honestly do.
Sebum swings enormously with time since cleansing and pH varies by site, so both are only read at fixed points in the loop, never opportunistically.
Adhesive-disc desquamation sampling and sebum-absorbent tape are long-standing dermatological research consumables, used here as published.
The swab
two sites · mails back
The living layer — sequenced at strain level, and for fungi separately, because abundance alone answers the wrong question.
The swab
- RTC. acnes ribotype balance — RT4/RT5 against RT6
- %Staphylococcus proportion, S. aureus resolved
- H′Shannon diversity of the bacterial community
- relMalassezia load, by ITS sequencing
Fitz-Gibbon and colleagues found the relative abundance of C. acnes was similar between acne and normal skin — only the strain population differed, with RT4 and RT5 associated with acne and RT6 with health. “How much C. acnes do you have” is the wrong question, and it is the one most tests answer.
These numbers are compositional — the sequencer imposes an arbitrary total, so one taxon rising forces the others down and no single abundance can be read alone. Sequencing cannot tell living from dead. And there is no validated healthy reference range: in atopic flares even the commensal S. epidermidis rises alongside S. aureus, so a tidy good-bug/bad-bug ratio would be a fiction.
Fitz-Gibbon et al., J Invest Dermatol 2013 (49 acne / 52 normal, nose follicles). Kong et al., Genome Res 2012 (temporal AD flares). Gloor et al., Front Microbiol 2017 on compositionality.
Dried blood spot
finger prick · mails back
Four systemic markers, including the honest measurable behind “skin age”.
Dried blood spot
- HbA1cGlycation — collagen cross-linking’s nearest routine proxy
- 25-OH-DVitamin D status
- hs-CRPLow-grade systemic inflammation
- FeFerritin
Lancet, spots onto filter paper, dried, posted. The format has been in continuous clinical use since newborn screening adopted it in the 1960s.
Haematocrit sets blood viscosity, which changes how far a drop spreads on paper — so a fixed punch from a high-haematocrit spot holds a different amount of blood than from a low one, and the resulting assay bias can exceed acceptance limits at values outside the normal range. We widen the interval we report to match, rather than quoting a precision the format does not have. Run by a CLIA-certified partner lab and reported as measurements, never diagnoses.
The haematocrit effect is the most widely discussed issue in quantitative DBS analysis; see the EBF DBS-Microsampling Consortium results and reviews in Bioanalysis, 2010–2014.
Dried urine, four points
one day · mails back
Steroid metabolites as a path across a day, never as a single number.
Dried urine, four points
- slopeCortisol from waking to night — four timed points
- 5αAndrogen metabolites: the sebum pathway
- 2:16Oestrogen metabolite ratio
- phaseCycle day, recorded with every sample
Four timed collections onto filter paper across one day, quantified by LC-MS/MS.
A single hormone reading in a cycling body is close to meaningless — phase and time of day dominate the value. Metabolites reflect clearance as well as production, so a high number has two causes and we say which we cannot separate. Oral contraceptives and HRT change the whole interpretation and are recorded as part of the reading.
Diurnal cortisol slope and urinary steroid-metabolite profiling are established endocrine methods; the four-point design follows standard diurnal sampling practice.
The patch
at home · twelve minutes
The only instrument here that measures dynamics instead of state — how fast you come back.
The patch
- stingSelf-rated 0–3 at 2.5, 5.0 and 8.0 minutes; mean of the three
- ΔEErythema peak, imaged against the card
- t½Recovery half-time, in minutes
- rateBarrier recovery, TEWL returning toward baseline
5% lactic acid to the nasolabial fold with saline on the contralateral side as the control, scored at 2.5, 5.0 and 8.0 minutes on a 0–3 scale — the protocol Frosch and Kligman published in 1977 and still the most widely used way to put a number on sensitive skin. Then the card and The Reader watch it resolve.
The sting score is self-rated and subjective, and we report it as exactly that rather than dressing it as an instrument reading. Recovery time shifts with room temperature. The patch is skipped entirely on active dermatitis, broken skin, or a recent procedure.
Every other instrument tells you where you are. This one tells you how much room you have left — and that is the number the system paces your routine against.
Frosch & Kligman, J Soc Cosmet Chem 1977 — the lactic acid sting test. Timings and scale as published.
The badge
worn · passive · the whole loop
The only instrument measuring what the world did to you, rather than what your body is.
The badge
- SEDCumulative erythemal dose across the loop
- SEDPeak single-day dose
- %RHMean ambient humidity
- °CAmbient temperature range
A cosine-corrected, erythemally-weighted electronic sensor logging time-stamped readings. 1 SED = 100 J/m² of erythemally effective UV, weighted by the action spectrum in ISO/CIE 17166. Published miniature dosimeters of this class resolve to 0.09 SED/h and agree with reference instruments to within 3% over a week.
A badge dose is not a facial dose. Received UV depends steeply on how a surface is oriented to the sun, so a lapel reading is a proxy for your face, not a substitute — we report it as exposure, never as dose-to-skin. It also reads ambient UV, so it cannot see what your sunscreen or your sleeve stopped. Consumer wearables differ measurably in spectral response, which is why this one is ours and calibrated rather than whatever is on your wrist.
It is the confound control. Without it there is no way to tell “the product changed my reading” from “it is July” — and loops sit weeks apart. Every other instrument measures you; this one measures the thing you were measured against.
Heydenreich & Wulf, Photochem Photobiol 2005 — miniature personal erythemal dosimeter, performance as quoted. ISO/CIE 17166 for the erythemal action spectrum and SED.
The breakthrough is not an exotic sensor. Every instrument here is published and most are decades old. It is that each one measures a loop at its own natural period — recovery over twelve minutes, the surface over two mornings, cortisol over one day, hormones over one cycle, exposure across all of them — and one geometry reads every clock.
Your Manifold
Confidence is whether the people near you agree — not how many there are.
Every reading places you somewhere in a twenty-seven-axis space. What decides whether we build for you is how tightly your neighbourhood holds together, axis by axis.
Example readings — three people, same apparatus
Bars show σ per axis. The hairline is τ — the ceiling a neighbourhood may not cross.
K = mean(σi) over the base
C = τ / K τ = 0.30
K = 0.170 → C = 0.30 / 0.170 = 1.76
Reading B is the case nobody else handles. Optically, at the surface, in her microbiome, her recovery and her exposure the neighbourhood is tight. On the three endocrine axes it scatters. So she gets her barrier routine built with confidence — and the androgen-pathway modifier held back, by name, with the reason printed.
Compare
What else is out there, fairly stated.
Every system below solves a real problem, and they are not all trying to do the same thing. Two of them do something ORÍ cannot do at all.
| System | How it decides | What it measures | Modalities | Declines when uncertain |
|---|---|---|---|---|
| ORÍthis page | Instrumented reading across 27 axes, then coherence over the neighbourhood | Optical · barrier · microbial · systemic · endocrine · dynamic · exposure | 7 | Yes — computed, and per axis. C = τ/K, with named holds |
| Muselytelehealth Rx | Online questionnaire plus three photos, reviewed by a board-certified dermatologist who writes a compounded prescription | Nothing instrumented — clinical judgement on images and history | 0 | Yes — a licensed prescriber can decline. Whole-case, not per-axis |
| Atollaacquired by Function of Beauty, 2021 | Monthly at-home test plus lifestyle and environment inputs through an app, into a recommendation algorithm | Skin moisture/hydration and pH — genuinely measured at home | 1 | Not published |
| Quiz-based customcategory | Self-reported questionnaire mapped to a formulation | Nothing. Self-report is not measurement | 0 | No |
| Clinic multispectral imagingcategory | Trained operator, fixed-geometry camera under controlled illumination | Optical only — pigment, redness, surface texture. Genuinely rigorous within that | 1 | n/a — it reports images; it does not formulate |
| DTC genetic skin testscategory | Genotype at a selected panel of variants | A fixed genotype — which does not change when your skin does | 1 | No |
Two honest notes. Musely’s model is clinical in a way ours is not — a licensed prescriber can prescribe drugs we cannot and will not, and a human declining to treat is a real and good form of refusal. And Atolla shipped at-home skin strips years before us, so we are not the first to measure at home. What is new here is seven modalities instead of one, loops instead of snapshots, and a refusal that is computed per axis rather than left unstated.
How it works
Five loops, five clocks.
Recovery
A tiny standardised provocation and the return from it. The shortest loop, and the one that sets the pace of everything else.
The surface
Bare morning, your day, bare morning again. You return to the same nominal state; your skin returns somewhere else. That gap is the reading.
Cortisol
Four dried-urine points from waking to night. A single cortisol value tells you almost nothing. The slope tells you a great deal.
Endocrine
Sampled across a cycle, or across four weeks where there isn’t one. Phase is recorded with every sample, so a value is never read naked.
Exposure
The badge logs underneath all of them, so a changed reading can be separated from a changed season.
One loop in every set runs on base only, no actives. If your gap does not collapse when the actives come out, the instrument is measuring your pillowcase — and we would rather find that out than sell past it.
A real reading
What your skin actually did.
Example reading, shown to illustrate the output.
The Kit
Plain formulas. Personal ordering.
The chemistry is deliberately ordinary and substantiated as a set. What is personal is which ones you get, in what order, on which days — and which ones we hold.
Every modifier names the axes it is gated on. If those axes scatter in your neighbourhood it does not ship, and the reason is printed on your reading rather than buried. If your gap grows or your recovery slows across loops — and the badge says the sun didn’t do it — the system removes a modifier. It can only ever take away.
Why this works
The industry’s model has no term for your skin.
Every claim about an ingredient “penetrating” traces back to a two-variable line fit on 93 compounds in 1992. Here is its error as a multiplier on the answer, rather than in log units.
That last bar is the whole idea. Narrowing the question cuts the error more than fivefold with no new chemistry. It has never been narrowed by measured skin, because nobody measured it. Twenty-seven axes is us measuring it.
Straight answers
What we won’t tell you.
There is no validated scale turning thickness, glycation and water loss into a number of years. We report the measurements. Everyone selling you a skin age invented the mapping.
No such reference range exists. We can tell you where you sit relative to people we have measured — never against a “normal” nobody has established.
Nobody can calculate that for an individual. The best published models carry a 7–8× error band and an at-home test does not fix it.
Measurements, not diagnoses — and no product here changes a hormone. We read your endocrine axes to decide what to withhold, never to claim we move them.
Sources
Every number on this page, and where it came from.
- Potts RO, Guy RH. Predicting skin permeability. Pharm Res 1992;9(5):663–669. DOI 10.1023/A:1015810312465. n=93, r²=0.67.
- VEGA QMRF, Skin permeation LogKp model, Istituto Mario Negri, Nov 2022, §3.7/§6.7. n=271, R²=0.40, RMSE 0.87 log units.
- Jachak A, Heckman C, Pagone F. J Environ Expo Assess 2025;4:42. DOI 10.20517/jeea.2025.42. Ten models, mean R²≈0.46, RMSE≈1.1; alcohol-stratified R²≈0.81, RMSE≈0.38.
- Waters L, Quah H. Sci Data 2022;9:584. RMSE 0.8→0.3 on fixing site, vehicle and temperature.
- Chardon A, Cretois I, Hourseau C. Skin colour typology and suntanning pathways. Int J Cosmet Sci 1991;13(4):191–208. ITA° = arctan[(L*−50)/b*]·180/π, six bands as used here.
- ITA° is computed from reflectance CIELAB, where L* is lightness and b* the yellow–blue axis, with the angle origin fixed at L*=50.
- Fitz-Gibbon S, et al. Propionibacterium acnes strain populations in the human skin microbiome associated with acne. J Invest Dermatol 2013;133(9):2152–2160. 49 acne / 52 normal; RT4 and RT5 acne-associated, RT6 health-associated; relative abundance similar between groups.
- Kong HH, et al. Temporal shifts in the skin microbiome associated with disease flares and treatment in children with atopic dermatitis. Genome Res 2012;22(5):850–859. Staphylococcus proportion rises in flares and tracks severity; S. epidermidis also rises.
- Gloor GB, Macklaim JM, Pawlowsky-Glahn V, Egozcue JJ. Microbiome datasets are compositional: and this is not optional. Front Microbiol 2017;8:2224. DOI 10.3389/fmicb.2017.02224.
- Frosch PJ, Kligman AM. A method for appraising the stinging capacity of topically applied substances. J Soc Cosmet Chem 1977;28:197–209. 5% lactic acid to the nasolabial fold; scored 0–3 at 2.5, 5.0 and 8.0 min; mean of three readings.
- Contemporary protocols commonly use saline on the contralateral site as control, as adopted here.
- Heydenreich J, Wulf HC. Miniature personal electronic UVR dosimeter with erythema response and time-stamped readings in a wristwatch. Photochem Photobiol 2005;81(5):1138–1144. Sensitivity 0.09 SED/h; within 3% of reference over 7 days; cosine-corrected.
- ISO/CIE 17166 — erythema reference action spectrum and standard erythema dose. 1 SED = 100 J/m² erythemally effective.
- Wearable consumer dosimeters differ measurably in spectral response; see August et al., Photochem Photobiol 2026, characterising three devices.
- The haematocrit effect is the most widely discussed issue in quantitative DBS analysis: haematocrit sets blood viscosity and therefore spot spreading, so a fixed punch holds different blood volumes at different haematocrits. See the EBF DBS-Microsampling Consortium results and reviews in Bioanalysis, 2010–2014.
- Musely — online questionnaire plus three photographs reviewed by a board-certified dermatologist, who writes a prescription filled as a compounded formulation by a partner pharmacy; visit fee stated by the company.
- Atolla — MIT-founded personalised skincare with a patented at-home test measuring skin moisture and pH, combined with lifestyle and environmental inputs in-app. Acquired by Function of Beauty, announced 4 August 2021.
- Remaining rows describe categories rather than named companies, and are characterised by method only.
- Coherence and chart admission are computed by GIGI. K is the mean per-dimension σ of a chart over the context base; a chart admits a point only if K stays at or below τ, default 0.30. C = τ/K.
- Example readings on this page are illustrative, chosen to show the mechanism. They are not measured results and are not presented as any person’s data.