16 million Americans have rosacea. The tracker category they need has been left structurally underbuilt.

The market that does not have its tool

There are roughly 16 million Americans diagnosed with rosacea, and that figure undercounts true prevalence because the diagnostic apparatus systematically misses cases (see our companion piece on rosacea in skin of color). The true population is larger; the diagnosed population alone is one of the largest single-condition cohorts in dermatology.

A chronic, relapsing, trigger-responsive, quality-of-life-affecting condition with 16 million American sufferers should have an obvious tracker category. The condition matches every property that drives sustained patient-tracking demand. The patient experience reliably includes years of trial-and-error trigger identification, multiple dermatology visits per year, and a constant low-grade question of whether the current management plan is actually working.

The tracker category that exists is small, structurally broken in consistent ways, and underbuilt. The rosacea-named apps available on iOS at this writing share a small set of design failures, none of which is a hard engineering problem to solve. The reason the category looks like this is not capability; it is that no one with the patience to read the literature and respond to it has built the tool. We are arguing that this is one of the largest indie opportunities in consumer dermatology, and that the structural shape of the underbuilt category is itself the prima facie case for building it properly.

Five structural failures the category ships

Pulling together the per-pillar pieces, the category fails in five consistent ways. Each is documented in the published dermatology literature, has been documented for years, and has not made it into the consumer-app implementations.

One: phenotype-incorrect onboarding. Most rosacea-named apps ask the user to pick a subtype at onboarding (erythematotelangiectatic, papulopustular, phymatous, ocular). The ROSCO 2017 framework formally retired the discrete-subtype model in favor of a phenotype-based approach (Tan et al. BJD 2017; Schaller et al. BJD 2020; Gallo et al. JAAD 2018). Patient-facing tools that still use the 2002 subtypes are nine years out of date relative to the field they purport to serve.

Two: collapsed-severity scoring. Most apps surface a single composite rosacea severity score (a 0-to-10 slider or a 0-to-4 grade) that collapses erythema, papules, telangiectasia, and other features into one number. The Wienholtz et al. 2023 JEADV paper validating the new RASI scale names this collapse as the structural failure mode of the IGA. A patient with severe erythema and no papules and a patient with moderate everything can get the same score; the resolution that matters for tracking progress is in the per-feature dimensions, and the global score discards it.

Three: ignored ocular phenotype. Ocular involvement affects 50 to 72 percent of cutaneous rosacea patients and often precedes the facial signs. The ROSCO 2017 framework lists it as one of four major phenotypes. Most consumer apps have no field for it in their daily log; the symptom that for many patients is the leading edge of their disease has no place to land in the tracker.

Four: no skin-tone equity work. Reported prevalence in Fitzpatrick IV through VI sits at 6 to 8 percent of the diagnosed population, with the literature explicit that much of the gap is diagnostic masking rather than biology. The dermatology image-classification model literature (Adamson and Smith JAMA Dermatol 2018; Daneshjou et al. Sci Adv 2022) documented 3 times performance degradation on darker skin in widely-deployed models. Most consumer skin apps surface a confidence number on a model that has not been validated on a balanced test set; the patient on Fitzpatrick V opens the app, gets a confident wrong assessment, and the diagnostic delay grows.

Five: broken clinical-photography pipelines. The published serial-comparability protocol (Nicholson 2022 JAAD; Lakhan 2022 BJD) requires manual white-balance lock, no flash, fixed framing, and consistent lighting. Most consumer apps inherit the iOS Camera framework defaults (auto WB, auto exposure, auto flash, free-form framing), which means their photo libraries are not serially comparable. The photos exist; the comparison signal does not.

Each of these five is a published problem with a published solution. The category-level pattern is that the consumer apps have implemented none of them, and the patient-facing literature has rarely surfaced the problems as problems.

Why the category has been left underbuilt

The honest question is why a $16-million-patient market has been left with apps that fail the published literature on five structural dimensions. Several reasons converge.

The first is that rosacea is unprestigious as a software market. The high-prestige dermatology software targets are melanoma detection (life-threatening, headline-grabbing), acne (large adolescent and young-adult cohort with high lifetime value), and aesthetic skincare (large discretionary spend). Rosacea is a chronic-management category with quieter conversion economics. It rewards the patient indie builder more than it rewards a venture-backed launch.

The second is that rosacea genuinely is medically complex enough to require literature-reading to build properly. An aesthetic-skincare app can be built without reading the dermatology literature; a rosacea tracker that does not read the literature ships the five failures above. The complexity is a moat but it is also a cost; the would-be builder who is not willing to put 200 hours into the medical-literature dossier ships the broken version, and the broken version is what the App Store has.

The third is that the dermatology and consumer-app communities have not crossed over enough. The literature on the phenotype framework, the skin-tone equity work, the photography protocols all exists in dermatology journals. The consumer-app builders mostly source their thinking from product reviews, marketing copy, and adjacent apps in their own category. The literature is the input that produces the right design; the consumer-app source set does not include it.

The fourth is that the patient communities (forums, subreddits, support groups) have learned to under-promise from their tools. Patients have downloaded several broken rosacea apps and gotten used to using them anyway, in the same way users learn to live with bad enterprise software. The demand signal that drives a better builder to enter is muted by the population having lowered its expectations.

None of these is a permanent structural barrier. They are reasons the category is the way it is, not reasons the category has to stay the way it is.

What the literature-grounded version looks like

If the five failures above are the negative space, what is the positive shape?

A phenotype-aware rosacea tracker built around the published dermatology consensus has the following design properties. We are describing this as a category prescription, not a list of Skinframe-specific features (Skinframe ships this shape because the literature recommends it; another builder could ship the same shape):

- No subtype question in onboarding. The patient turns on phenotype-feature toggles (persistent redness, transient flushing, papules and pustules, visible vessels, ocular involvement, burning and stinging) as the features show up. The data model treats features independently, per the ROSCO 2017 phenotype framework. - Per-feature severity scoring, with PSA-style descriptors (clear, almost clear, mild, moderate, severe) for the erythema dimension and appropriate scales for the other dimensions. No single composite score; each feature on its own line. - Ocular log on the first daily-log surface, not buried two settings deep. The ocular dimension is treated as a major phenotype because the literature treats it as one. - Skin-tone-branched UX. Fitzpatrick onboarding picker; Fitzpatrick IV through VI users see the sensory diary foregrounded; the photo-derived redness signal is not surfaced as a top-line trend until the photo handling is validated for darker skin. The published bias literature (Adamson and Smith 2018; Daneshjou 2022) is named on the methodology page. - Clinical-photography pipeline. Manual WB and exposure lock at first capture, replayed across sessions. Flash always off. Distance and angle enforced via Vision and CoreMotion. Onboarding lighting and background guidance, including the dark-blue-background recommendation for skin of color. - Per-patient baselines, not population-trained classifiers. Each user is treated as a sample of one. No diagnostic inference from photos. The patient owns the trajectory of their own dimensions. - Three-tier evidence-labeled trigger taxonomy. Tier A canonical triggers ship as default chips with one-line evidence notes. Tier B emerging or individual-variable triggers (hormonal cycle, post-COVID, sleep deprivation) are opt-in. Tier C is user-defined free-text. - Single-page derm-handoff PDF, with three top-band numbers (flare count, average per-feature severity, quality of life trend), trigger co-occurrence with evidence tiers, two photos (good day and bad day), and a methodology footer citing the phenotype framework. PDF first, CSV second, FHIR for the practices that ask. - Privacy posture. Photos in the user's iCloud, not the developer's servers. Editorial integrity bar on every clinical claim. No diagnostic inference. The user is in charge of their own data.

This is not a long feature list; it is a literature-grounded design. The category has had it as a possible design for at least nine years, since ROSCO 2017. It has not had a shipped tool implementing it.

Why an indie is the right shape of builder

The rosacea-tracker category is the kind of indie-builder opportunity that maps poorly to venture-backed product economics and well to a patient builder who reads the literature.

The venture-backed dermatology software companies (the ones that have raised meaningful capital) have largely targeted higher-volume conditions and higher-prestige indications. Rosacea has been a footnote in their roadmaps. A venture-backed entrant would face the same opportunity-cost calculation those existing companies made, would size the market against the visible diagnosed cohort rather than the true prevalence, and would arrive at the same conclusion: not their fight.

The indie builder, working with consumer-app economics (no investor return target, no growth-at-all-costs pressure, no need to acquire the entire market in 18 months), can sustainably build a tool for the 16 million-plus rosacea cohort with the right design and the right voice and ship it as a lifetime-priced product. The unit economics work for the indie; they would not work for the venture-backed alternative. This is the patient-indie thesis applied to rosacea specifically.

The trade is that the indie builder has to do the work the dermatology literature requires. There is no shortcut. The medical-literature dossier we work from is 4,451 words and 15 primary sources; the design choices follow from the reading. The indie builder is unusually well-positioned to do that reading carefully, because they are accountable to the patient using the tool, not to a quarterly forecast that does not have time for the literature.

Skinframe is what falls out the other end of this argument. We are the indie answer to a structurally-failed category that the larger software market has decided not to fix.

What this article is trying to do

Two things, and we will be straight about both.

The first is that this is the most-likely-shareable piece in our editorial set. The category-thesis frame travels: it works for a dermatologist reading their patients' tool options, for a clinically-literate patient evaluating whether to switch trackers, for a tech-curious reader interested in why some markets get left underbuilt, and for the indie-software community as a worked example of a literature-grounded design.

The second is that this piece exists for the reader who has already read several of our per-pillar pieces and wants the synthesis. The five-structural-failures framing only lands if the reader has either seen the per-pillar arguments or is willing to take them on the references. We have linked the per-pillar pieces above; the structure of the editorial set is that each pillar argues a specific position with the literature, and this synthesis names the position the per-pillar arguments collectively support.

What we are not trying to do here is sell the app, though we will name it. Skinframe is the implementation of the design we are describing. It is shaped this way because the literature recommends it. If a reader walks away convinced that the literature-grounded design is the right design and chooses a different builder's implementation, the literature still wins, and the patient still gets the tool they should have. That outcome is fine. What we are arguing against is the version of the rosacea-tracker category that has been the default for the past decade, which fails the published consensus on five structural dimensions and has shipped to the App Store anyway.