For a long time, glow was sold as a thing you could buy. A serum, a 12-step ritual, a single hero product with a viral name. Spend enough, layer enough, and the glow would arrive.
It was never true. The glow everyone is actually chasing is not a product and not a finish. It is a symptom — the visible sign of skin that is healthy: calm, even, well-hydrated, protected, and behaving the way your skin behaves when it is well. You cannot bottle that, because it is not one formula. It is the right things, for your skin, done consistently. Which means the real luxury was never the product. It was the personalization. Healthy skin, built for you — that is the new glow.
And if that is the standard, then one group of people has been quietly cheated of it for decades.
The market clusters at one end of the gradient. Zamia is built for the whole thing.
1. What this paper rests on, and what its companion paper covers instead
This is the earlier of two Flux papers on this subject. A later paper grounds the clinical and regulatory case in depth — the dermatology literature on skin-of-color underrepresentation in trials, textbooks, and AI training data, and the mechanism and persistence of post-inflammatory hyperpigmentation (PIH) specifically. Having read that paper in full before writing this one, the division of labour between them is deliberate: this essay does not quietly repeat its apparatus.
What this paper argues from. The lived, repeatedly reported experience of trial-and-error that melanin-rich skincare customers describe, which I treat as a pattern worth taking seriously rather than as measured incidence data. The documented history of the beauty industry's shade-range expansion, specifically Fenty Beauty's 2017 launch and an independent data-journalism analysis of what the market's shelves actually looked like before and after it. Peer-reviewed and press-reported figures on skin-bleaching prevalence in African markets specifically — a geography the later paper's global WHO/FDA/EU sourcing does not focus on. And, for the single empirical clinical claim this essay still makes directly — that PIH can persist for months after the irritation that caused it — the same primary source the later paper verified, cited here rather than re-derived.
What this paper does not argue from. It does not argue from a clinical trial, a dermatological study Flux commissioned, or any audit of Zamia's own evidence engine; the later paper's own §7 states plainly that such an audit has not been done, and nothing in this essay should be read as implying otherwise.
What this paper does not claim. It does not claim mainstream skincare brands deliberately excluded darker skin out of malice — §4 is explicit that the shade gap and the clinical-evidence gap both trace to who was cheapest and easiest to recruit and photograph, a structural pattern rather than a conspiracy. It does not claim bleaching products are used only out of ignorance; §5 treats the motivations the literature actually reports — including social and economic pressure, not only misinformation — honestly.
What would refute it. In decreasing order of damage, given in full in §9. The sharpest version: if melanin-rich skincare customers turn out not to experience materially more product mismatches, more marking complications, or a longer path to a working routine than lighter-skinned customers do — if the "trial-and-error tax" this essay is built on turns out to be a story rather than a pattern — the whole argument for why personalization specifically matters more here collapses, and ordinary catalog expansion (more shades, more product lines) would be the right response instead of what Zamia is built to do.
2. The trial-and-error tax
If you have deep, melanin-rich skin, you already know the routine that is not on any label. You buy the product everyone swears by. You use it. It does nothing, or it stings, or worse — it leaves a dark mark exactly where the breakout was. You stop. You research for an hour, half of it contradictory, none of it clearly about your skin. You buy the next thing. Repeat.
That is the trial-and-error tax, and it is enormous. It is paid in money — a shelf of half-used bottles that were never going to work. It is paid in time — months of guessing per actual result. And it is paid in the skin itself, because for this skin, a bad guess does not just fail quietly. It marks: a product that pushes the skin too hard can trigger post-inflammatory hyperpigmentation, a pigmented mark that the clinical literature documents can take months to resolve even once treated, and longer — sometimes effectively indefinitely — when the pigment migrates deeper into the skin untreated.1 I am not going to re-derive the mechanism here; the later paper does that in full, with the dermatology literature behind it. What matters for this essay is narrower: the cost of a wrong guess is not symmetric across skin types, and a routine built without reference to that asymmetry is gambling with a cost it never prices in.
Nobody chose this. It is the cost of being handed products built for someone else and being told to figure out the rest yourself. I do not have a verified figure for the average money or time this costs an individual customer — no study measuring it was found, and I am flagging that in Appendix B rather than estimating a number to fill the gap.
3. Built for someone else, sold to everyone
Here is the uncomfortable engine underneath it. Most skincare is formulated, tested, and photographed on lighter skin. The clinical images, the trial panels, the "before and afters," the influencer demos — the default face is a lighter one, and the science was tuned to it. Then the product is sold to everyone as if skin were one thing.
It is not one thing. Melanin-rich skin behaves differently in the way that matters most: it responds to irritation by producing pigment, not merely by reddening and recovering. Push it with a too-strong acid, an aggressive scrub, a product that wasn't right, and the failure mode is not a day of redness. It is the pigmented mark described in §2. So the loud, generic playbook ("use the strongest thing, push through the purge") is not merely unhelpful for this skin. It is backwards. It manufactures the exact problem people came to solve.
The market's answer to all of this, for years, was even worse, and §5 takes up that history directly. So deep-skinned people were left in a gap with three bad options — products not made for them, advice not about them, or an industry telling them the problem was their skin tone itself. The trial and error wasn't a personal failing. It was the only path left when nothing was built for you.
4. Representation on the shelf, before anyone called it that
It is worth being precise about what actually changed in the beauty industry's shade ranges, and when, because the popular story ("there just weren't enough shades") is not quite what an independent look at the shelf found.
Rihanna's Fenty Beauty launched its Pro Filt'r foundation in September 2017 with 40 shades spanning the full range from very light to very deep, and the launch was widely covered as a genuine industry turning point — named to TIME's "25 Best Inventions of 2017" specifically for that range.2 But a data-journalism analysis published the following year, mapping the actual shade composition of major brands' foundation lines, found that the number forty was not new: Maybelline had already expanded to 40 shades months earlier, in May 2017, and Make Up For Ever had offered 40 shades since 2015.3 What was new was the distribution. Of Make Up For Ever's 40 shades, the same analysis found 31 clustered in the lightest third of the lightness scale measured — a shade count that looked inclusive on a marketing page and was not, in practice, inclusive on a shelf.3
That distinction matters for this essay's argument in a way it did not for the industry's own self-congratulation at the time. A brand can expand its shade count, generate a press cycle, and still leave the exact customer this essay is about holding a tube that is close but not right — which is a segmentation improvement (more buckets to sort people into) rather than the deeper kind of personalization this essay is arguing for, one that accounts for how a given customer's skin specifically responds rather than just what color it starts as. The later paper gives that distinction a name — segmentation-personalization versus risk-personalization — and shows it applies with even higher stakes to the clinical-evidence gap than to the shelf itself. I am not re-deriving that argument here; I am noting that the shelf told the same story first, in its own more visible register, years before either paper was written.
5. The bad answer the industry already tried
For decades, the market's dominant response to "skin of color has different needs" was not personalization, in either sense. It was brightening — a euphemism, in a meaningful share of cases, for lightening or bleaching. The later paper documents the global regulatory history of this in detail — the WHO's warnings on mercury-containing lightening products, the FDA's and EU's bans on hydroquinone, the enforcement-testing figures on how much non-compliant product still circulates.4 I am not repeating that apparatus here. What this essay adds is a geography that global regulatory sourcing does not center: how widely this practice is actually reported within African markets specifically, since Flux's own customers are disproportionately there.
A 2023 short communication in the International Journal of Surgery, drawing together country-level studies across the continent, reports regularly-used skin-whitening products among roughly 25-80% of women across surveyed African countries, with individual country figures cited at 77% in Nigeria, 59% in Togo, 35% in South Africa, and 25% in Mali.5 The same piece attributes a widely repeated claim — that over 40% of African women bleach their skin — to a 2011 World Health Organization report; I have not independently fetched that 2011 document, and I am stating the attribution rather than the figure itself as verified, because that is the honest description of what I actually checked. The motivations the literature reports are not simply "misinformation": alongside cosmetic concerns (clearing blemishes, evening tone), it names social pressure around lighter skin and attractiveness, relationship dynamics, and a longer colonial history shaping whose skin tone gets treated as the default — a more complicated picture than "people just don't know the risks," and one this essay is not going to flatten for the sake of a cleaner argument.5
The reason this matters for the personalization argument, rather than being a separate topic, is this: bleaching was never personalization, and it was never trying to be. It was a single, universal intervention — reduce the melanin — applied regardless of what any individual's skin actually needed, marketed as care. The correct response to "the mainstream market underserves darker skin" is not "sell darker skin a different universal product." It's the harder, more specific thing: build recommendations that account for how darker skin actually responds to irritation, without ever treating the pigmentation itself as the thing to be corrected.
6. Personalization is the answer — and now it scales
The fix is almost embarrassingly simple to state and was, until recently, impossible to deliver at scale: stop selling everyone the same thing and start with the actual skin in front of you.
What does your skin do? What does a breakout leave behind on it? What have you already tried, and how did it actually go? What is your routine realistically going to survive? A great consultant — a dermatologist or esthetician who genuinely understands melanin-rich skin — starts there, then builds a small, honest routine of real products chosen for real reasons, and tells you what to skip. That ends the trial and error. It replaces guessing with a plan.
The catch was always access. There were never enough of those experts, and the ones who exist are expensive and far away. So the knowledge existed; the person who needed it could not reach it. That gap is precisely what consumer AI can now close — not because a model "knows skincare," but because a well-built one can do what a great consultant does: listen, ask one sharp question instead of ten generic ones, remember what you already tried, and respond like someone who respects your intelligence.
It is worth naming, honestly, how unsettled even the commercial side of this claim is. Two market-research publishers, nominally sizing the same "personalized skin care" category for the same base year, 2025, put the figure at USD 26.6 billion and USD 33.06 billion respectively — a roughly 25% gap between two firms claiming to measure the same thing.6 I am not citing either number as a reliable estimate of the opportunity; I am citing the gap between them as evidence for something this essay has been arguing on other grounds all along — that "personalized skincare" is not yet a category anyone, including the analysts paid to size it, agrees on the boundaries of.
7. Zamia
That is what Zamia is. A personal skincare consultant for Black and African women — and for anyone with deep, melanin-rich skin — that turns "buy it and hope" into a routine built for you.
She is built around one decision everything else follows from: she is boxed. Zamia leads the conversation in her own warm, plain-spoken voice — empathy first, then the honest next step — but she does not invent. Every hard fact — the actual products, the prices, the routine itself — is supplied by a vetted engine and shown to you as cards. She phrases, reassures, and decides; the catalog and the evidence sit underneath her, fixed. The difference between a chatbot that will cheerfully invent a product and a consultant whose every recommendation is real and chosen for a reason she can name — that difference is the whole product. Trust is the entire game, so she would rather ask one more question than guess.
And she holds one line that is not negotiable: radiance, never lightness. She helps skin become clearer, calmer, more even, more itself — never lighter. She warns against bleaching and unsafe brightening, on purpose, in code. Because the goal was never to change the skin you have. It was to make it healthy.
Whether Zamia's underlying evidence engine is, in practice, curated against skin-of-color-specific sources rather than the same generically lopsided dermatology literature §3 and the later paper both describe is an empirical claim about Flux's own product that neither paper has independently audited. I am stating that limitation here, not only in the later paper, because this is the essay that makes Zamia's case to a general reader first.
8. Objections I take seriously
"Flux has a commercial interest in this conclusion." Flux builds and sells Zamia, and this essay exists in significant part to make the case for why Zamia should exist. That is a direct financial stake in readers accepting that risk-personalization — not just more shades on a shelf — is the thing actually missing, and that Zamia is a serious attempt at it. The honest mitigation, as the later paper also states, is narrow: §§4-5's representation history and prevalence data come from independent journalism and peer-reviewed literature that would be true whether or not Zamia existed. Whether Zamia's own engine lives up to what §7 describes is Flux's claim to make and the reader's to doubt.
"The shelf has already fixed this — Fenty proved the market responds." §4 takes this seriously rather than dismissing it: shade ranges genuinely expanded industry-wide after 2017. But The Pudding's own finding — that two brands already had 40-shade ranges before Fenty, skewed toward light skin — is itself the rebuttal: a wider shelf is not the same achievement as a shelf whose breadth actually reaches the customer standing in front of it. More shades address the segmentation problem. They do not, on their own, address whether any given product in that range was tested against this skin's specific irritation response, which is the risk-personalization problem §3 and the later paper are both about.
"Framing skin bleaching as purely a market failure ignores real social and economic pressures that a better product can't fix." This is a fair challenge to this essay's own optimism. §5 states plainly that the literature's reported motivations include social pressure around attractiveness and a colonial history of valuing lighter skin — causes no skincare consultant, AI or human, can resolve by itself. Zamia's "radiance, never lightness" design can refuse to supply a bleaching recommendation; it cannot, on its own, change why someone might still want one. This essay's claim is narrower than solving that: it is that whatever the reasons someone seeks a lighter result, a tool that will only ever point toward healthier skin rather than toward lightening is a better thing for that tool to be than one that is agnostic between the two.
"This essay and its companion paper are the same argument twice, dressed up as two papers." I considered this seriously enough to read the companion paper in full before writing a word of this essay, specifically to check it. They are not the same claim: this essay argues that personalization — in the general, structural sense — is what "glow" always meant, using market and representation history; the later paper argues a narrower, falsifiable claim about one specific clinical risk (PIH) and the specific architecture (boxing a model against a vetted evidence engine) needed to price it in. A reader could accept this essay's claim and reject the later paper's, or vice versa, which is the actual test of whether two papers are making one argument or two.
9. What this paper leaves unquantified or unaudited
The "trial-and-error tax" in §2 is not quantified in money or time: no study measuring it was found, and Appendix B registers that gap rather than estimating a figure to fill it.
It does not claim Zamia's evidence engine has been independently audited for skin-of-color-specific sourcing; §7 and the later paper's own §7 both say this directly.
It does not claim skin bleaching is practiced only out of misinformation; §5 reports social, relational, and historical motivations alongside cosmetic ones, on the literature's own terms.
It does not claim the 2011 WHO figure often repeated for African skin-bleaching prevalence (">40% of African women") as independently verified; §5 states plainly that this paper checked the citing literature, not the original WHO document.
And it does not claim this essay and its companion paper are testing the same hypothesis; §8's fourth objection states why they are not, and a reader is free to find one more persuasive than the other.
10. What would falsify this
Stated as commitments, in decreasing order of how much damage each would do to this paper's claims.
Melanin-rich skincare customers are not shown to experience a measurably higher rate of product mismatch, marking, or routine abandonment than lighter-skinned customers. If a well-sampled study found no material difference in trial-and-error burden by skin tone, the premise this entire essay is built on — that personalization matters more, not just differently, for this group — would fail, and the right response would be withdrawal, not a narrower restatement.
The shade-range expansion documented in §4 is shown to have closed the risk-personalization gap, not just the segmentation gap. If independent testing found that today's wide shade ranges are matched to formulations that account for differential irritation response by skin type — not just color — the distinction this essay draws between a wider shelf and a genuinely personalized one would collapse, and ordinary catalog expansion would be a sufficient response after all.
Zamia's own evidence engine is found, on independent audit, to be built on the same generically skewed dermatology sourcing this essay criticizes. This would not falsify §§2-5's account of the market-wide problem, but it would falsify §7's implicit claim that Zamia is a serious answer to it rather than a repackaging of the same gap under a friendlier interface.
A rigorous survey finds skin-bleaching practice in African markets driven overwhelmingly by factors a skincare consultant — AI or human — cannot meaningfully influence, such that no change in product personalization moves the underlying behavior at all. This would not falsify the market-representation argument in §§3-4, but it would substantially narrow any claim that Zamia's existence changes anything about §5's historical pattern specifically.
Appendix A — Evidence table
| # | Claim used in this paper | Value / finding | Source |
|---|---|---|---|
| 1 | PIH can persist for a protracted period, in some cases indefinitely, if untreated | Epidermal PIH: months to years untreated; dermal PIH: can be effectively permanent | Davis & Callender 2010, also cited in WP411 |
| 2 | Fenty Beauty's 2017 launch and industry recognition | 40-shade range, September 2017; named to TIME's 25 Best Inventions of 2017 | Contemporaneous press, multiple independent accounts2 |
| 3 | Shade-range quantity existed before Fenty but skewed light | Maybelline 40 shades (May 2017); Make Up For Ever 40 shades since 2015, 31/40 in the lightest third | The Pudding, "The Diversity of Makeup Shades" (2018), data-journalism analysis3 |
| 4 | Skin-bleaching prevalence in African markets | 25-80% range across countries surveyed; Nigeria 77%, Togo 59%, South Africa 35%, Mali 25% | Adetoogun et al. 2023, Int. J. Surgery, doi:10.1097/JS9.00000000000001415 |
| 5 | ">40% of African women bleach their skin" | Attributed to a 2011 WHO report by the citing literature; original WHO document not independently fetched by this paper | Adetoogun et al. 2023, citing WHO 2011 (unverified at source)5 |
| 6 | Two market-research firms' 2025 size estimates for "personalized skin care" diverge by ~25% | USD 26.6B (GMI) vs. USD 33.06B (Precedence Research), same nominal category, same base year | Global Market Insights; Precedence Research6 |
| 7 | Average money or time cost of the "trial-and-error tax" to an individual customer | — | — |
| 8 | Independent audit of Zamia's evidence-engine source composition | — | — |
| 9 | Measured difference in product-mismatch or routine-abandonment rate by skin tone | — | — |
Appendix B — Citation register
CITATION-NEEDED — the trial-and-error tax, quantified. Appendix A row 7. No study was found measuring the average money or time a melanin-rich skincare customer spends on product mismatches before finding a working routine; this paper's §2 describes the pattern qualitatively and does not estimate a figure.
CITATION-NEEDED — Zamia's evidence-engine composition. Appendix A row 8. As the later paper's own §7 and Appendix B also state: this paper asserts Zamia is designed to draw on skin-of-color-specific evidence but has not independently audited the engine's actual source composition. Carried forward here rather than re-opened as if it were a new gap.
CITATION-NEEDED — measured difference in mismatch/abandonment rate by skin tone. Appendix A row 9. §9's falsification condition names this directly: no study comparing trial-and-error burden by skin tone was found, and this essay's central premise rests on the pattern being real without a measured incidence rate behind it.
Davis EC, Callender VD. "Postinflammatory Hyperpigmentation: A Review of the Epidemiology, Clinical Features, and Treatment Options in Skin of Color." Journal of Clinical and Aesthetic Dermatology. 2010;3(7):20-31. Independently verified for WP41; cited directly here for this essay's single clinical claim rather than re-derived. See WP41 for the fuller mechanism and epidemiology.↩
Contemporaneous press coverage of Fenty Beauty's Pro Filt'r launch, September 2017, and its inclusion on TIME's "25 Best Inventions of 2017" list, cross-checked across multiple independent outlets.↩
The Pudding, "The Diversity of Makeup Shades" (June 2018), pudding.cool/2018/06/makeup-shades/. Fetched and read directly. A data-journalism analysis, not a peer-reviewed study; treated accordingly.↩
See WP41 §5 for the full, independently verified regulatory history (WHO mercury guidance, FDA and EU hydroquinone restrictions, EDQM enforcement-testing figures) — not repeated in this essay.↩
Adetoogun JI, Aderinto N, Ashimi AA, Akano DF, Ogundipe TO, Fikayomi PB. "Practice and motivations for skin bleaching among Africans." International Journal of Surgery 109 (2023): 218-219, doi:10.1097/JS9.0000000000000141. Fetched and read directly (PMC open access). Its attribution of a ">40% of African women" figure to a 2011 WHO report is reported here as an attribution, not as independently verified against the original WHO document.↩
Global Market Insights, "Personalized/Customized Skin Care Market" (2025 edition), gminsights.com — USD 26.6 billion (2025), USD 50.9 billion (2035 projection). Precedence Research, "Personalized Skin Care Products Market" press release (2025) — USD 33.06 billion (2025), USD 61.61 billion (2033 projection). Both fetched and read directly; cited for the divergence between them, not as a reliable market-size estimate in their own right.↩
