What topical skincare can and cannot do

Published 6 August 2026Zane Hitchcox, editor

The argument on this page is not that skincare is useless. It is that the useful things it does are a much shorter and much more boring list than the front of a jar suggests, and that the boundary between the two is not a matter of opinion. It is set by a piece of tissue about as thick as a sheet of kitchen foil.

The barrier is the ceiling, and it is a good barrier

Skin's outermost layer, the stratum corneum, is described in the delivery literature as a brick-and-mortar structure: the bricks are dead, flattened corneocytes packed with cross-linked keratin, and the mortar is a densely ordered lipid matrix arranged largely in bilayers.¹ The layer is 10 to 20 µm thick.¹ A 2022 review of topical and transdermal delivery describes the same arrangement — terminally differentiated cells embedded in tightly packed lipid lamellae, "in a structural arrangement analogous to a 'brick and mortar wall'".²

The mortar is not ordinary cell membrane. The intercellular lipids that build those bilayers are ceramides, cholesterol and fatty acids, with ceramides the major constituent, and it is the loss of exactly those lipids that damages the water barrier and produces dry skin.⁵ The surface is also acidic: a multicentre study of 330 people measured a mean skin surface pH of 5.12, falling to 4.93 after a day without washing or applying anything, from which the authors estimated a natural surface pH of about 4.7.⁴ The layer is not static, either — corneocytes are continuously shed from the top and replaced from below, which is why anything that merely sits on the surface has a short tenancy.

All of this is worth stating plainly because the marketing frame runs the other way. A great deal of cosmetic copy treats the barrier as damage to be overcome — something to be "penetrated", "delivered through" or "repaired" — when in healthy skin it is functioning equipment doing the job that keeps a person from desiccating. A product that genuinely got large molecules through it at scale would not be a cosmetic. It would be a drug delivery system, and it would be regulated as one.

There are three routes in, and one of them does most of the work

Passive movement across the stratum corneum happens by three routes: intercellular, winding between the lipid lamellae; transcellular, straight through the corneocytes and the lipid matrix in turn; and trans-appendageal, down hair follicles and sweat ducts.² In practice, transport across the layer typically means diffusion through the intercellular lipids, along a path that winds tortuously around the corneocytes.¹ The follicular route is real and is of live research interest, particularly for particles, but appendages occupy a small share of the skin surface and are not the main road.

What survives that journey has a recognisable profile. Successful transdermal drugs have molecular masses of only up to a few hundred daltons, partition coefficients that heavily favour lipids, and doses measured in milligrams per day or less.¹ The 2022 review puts the rules of thumb more explicitly: an optimal octanol–water log P between 1 and 3, a relatively low molecular mass under 500 Da, and a low melting point.² A molecule that is too water-loving will not enter the lipid mortar; one that is too oil-loving will enter it and stay there rather than moving on into living tissue.

The 500-dalton figure is a rule of thumb from one paper, not a law

The number gets quoted as though it were a constant of nature. It is not. It comes from a 2000 paper in Experimental Dermatology by Bos and Meinardi at the University of Amsterdam, which proposed that a compound's molecular weight must be under 500 daltons to allow skin absorption, and argued the case from three observations: that virtually all common contact allergens are under that weight, that the topical dermatological agents in ordinary clinical use are under it, and that all then-known transdermal drugs were under it.³ The paper's own recommendation was to restrict development of new topical compounds to that range.³

That is an argument from the observed distribution of things that work, not a measured cut-off, and it should be read as a heuristic with soft edges. Absorption falls off steeply with size rather than stopping at a threshold; formulation, occlusion, the vehicle, damaged or inflamed skin and site of application all move the line. But the heuristic is robust enough to do the job it is used for here. Any claim that a large molecule crosses intact skin in useful amounts is a claim against a strong prior, and the burden of proof sits with the claim.

What clears the bar: water, light and a short list of small molecules

Three categories of topical do work, and it is worth being exact about what "work" means in each.

Moisturisers work on the surface, which is where the problem is. The standard taxonomy divides them into occlusives, which physically block water loss by laying a hydrophobic layer over the stratum corneum; humectants, which are hygroscopic and draw water in; and emollients, lipids and oils that improve softness and flexibility.⁵ The size ordering is stark: that review reports petrolatum at a minimum concentration of 5% reducing transepidermal water loss by more than 98%, and lanolin, mineral oil and silicones reducing it by 20–30%.⁵ Glycerol is described in the same review as the most effective humectant.⁵ None of this requires anything to penetrate anywhere, which is precisely why it is the best-established thing a cosmetic does. The largest body of controlled evidence for moisturisers comes from eczema rather than from cosmetic use — Cochrane's 2017 review pooled 77 randomised trials and 6,603 participants — and it found benefit while grading the certainty outcome by outcome, from very low to high.⁶

Sunscreen is the best-evidenced preventive topical there is, and the evidence is for prevention over years rather than improvement over weeks. The Nambour trial in Queensland randomised community adults to daily versus discretionary sunscreen and followed skin microtopography for four and a half years, with blinded assessment.⁷ It is not a cosmetic result; it is the one place where an ordinary bathroom product has been shown in a randomised trial to slow visible skin ageing. What the SPF number on the front actually certifies, and why New Zealand made that testing mandatory, is a separate subject covered in the article on SPF and the New Zealand sunscreen standard.

A short list of actives changes the living skin below. Retinoids are the clearest case: they are small, lipophilic and have been through decades of vehicle-controlled trials in photodamaged skin. A 2025 systematic review and meta-analysis pooled eight randomised trials of tretinoin and found consistent but modest improvements against vehicle, along with a threefold increase in the odds of an adverse event — dryness, erythema, peeling, burning and stinging, mostly in the early weeks.⁸ Vitamin C and niacinamide belong in the same conversation with different evidence bases; all three are examined in detail in the article on actives with real evidence.

Note what these three have in common. The effect sizes are real and small, the outcomes are measured on graded scales rather than transformations, and every one of them took months. Nothing in this category behaves like the before-and-after photograph.

Botanical extracts mostly work less than the label implies

The commonest gap between promise and physics is not fraud. It is concentration. Ingredient lists are ordered by weight only down to one percent: under the United States labelling rule, ingredients present at a concentration not exceeding 1% may be listed in any order after those present at more than 1%, and colour additives in any order after that.⁹ Below that line the sequence carries no information at all, so a celebrated plant extract can sit near the top of the tail and still be present at a fraction of a percent, whatever the front of the pack implies. That mechanic, and how to read it, is set out in the guide to reading an INCI list.

Two things follow. First, an extract studied at a laboratory concentration in a dish tells you very little about the same extract at 0.1% in a lotion, because the exposure is not comparable and the barrier was not in the dish. Second, "less than advertised" is not "nothing": plant oils are emollients and behave as emollients regardless of provenance, and some botanical constituents are small enough to cross. The specific New Zealand ingredients this address is associated with are worked through one at a time in the botanicals reference, and for several of them the honest certainty label is not established.

Topical collagen is the cleanest example of a claim that cannot work as stated

Collagen is worth singling out because the arithmetic is unusually clean. A single type I collagen triple helix is a rod under 2 nm in diameter and roughly 300 nm long.¹⁰ A 2025 review of collagen hydrolysates in cosmetic formulation puts native collagen at "a high molecular weight of around 300–400 kDa", which is why it is insoluble in water and oils.¹¹ Set that against a 500-dalton rule of thumb³ and the conclusion is not marginal: intact collagen applied to skin is several orders of magnitude too large to cross the stratum corneum, and a jar of it cannot be replacing the collagen in anyone's dermis.

The fair version of the story is more interesting than the debunk. Cosmetic collagen is usually hydrolysed — broken by chemical or enzymatic treatment into smaller peptides, typically up to about 10 kDa, with the fish hydrolysate in that review averaging 7.5 kDa.¹¹ Fragments below 500 Da are, the same review notes, well absorbed, unlike native collagen.¹¹ But most of a typical hydrolysate is nowhere near that small, and being absorbed is not the same as arriving somewhere useful and doing something measurable once there.

What collagen and its hydrolysates demonstrably do is act on the surface. They have excellent water-binding capacity, and the resulting occlusion reduces transepidermal water loss and forms a protective film on the skin.¹¹ In other words, topical collagen is a decent humectant and film-former sold as a structural repair. The product is not inert and the customer is not defrauded of all value; the mechanism on the box is simply not the mechanism in the jar.

The label rules do not require the claim to be true

Two regulators, two ways of drawing the same line, and neither of them audits efficacy the way readers assume.

The practical consequence is the same on both sides of the Pacific. The rules police composition, safety and wording. They do not require a manufacturer to show that a cosmetic claim is true before the product is sold, which is exactly why a page like this one has to go to the trial literature instead of to the packaging. How a claim gets onto this site, and what the certainty labels mean, is set out in the method.

Past the barrier, the intervention stops being a product

The honest boundary is this: the treatments that reliably act below the stratum corneum do so by breaching it — with light, needles, acid at professional depth, or a syringe. That is a categorical change, not a stronger version of skincare. It brings a clinician, a consent conversation, downtime, a risk of scarring, infection and post-inflammatory pigmentation, and a bill.

It does not bring better evidence. Cochrane's review of interventions for acne scars — 24 randomised trials, 789 adults — found moderate-quality evidence favouring injectable fillers over placebo, and broadly comparable results between fractional laser, chemical peeling, fractional radiofrequency and skin needling. Almost everything else in it is graded very low quality, often on a single small study. Its overall verdict was blunt: there is a lack of high-quality evidence because of poor methodology, underpowered studies and inconsistent outcome measures, and the results "do not provide support for the first-line use of any intervention".¹⁴ The review also records that chemical peeling was not tolerable for 7 of the 43 participants who received it, or 16%.¹⁴

Microneedling is the procedure closest to the consumer end of this spectrum, and the pattern repeats. A 2021 systematic review in the International Wound Journal collected nine randomised controlled trials of microneedling for atrophic scars, found consistent efficacy signals with no serious adverse effects reported, and concluded that larger trials with longer follow-up are still needed to validate it.¹⁵ That is a defensible reason to consider a procedure. It is not the same standard of proof as the sunscreen trial above, and a reader deciding between a serum and a course of treatment deserves to know which claim rests on what.

The other half of that comparison is cost, and this is a place where an article can easily mislead by inventing a range. Prices vary by country, city, operator and device, and a made-up figure would be worse than none. Most clinics also decline to publish any figure at all, reserving it for a consultation, which is what makes the ones that do publish worth citing. Sarah Hitchcox Aesthetics, a med spa in the Knoxville, Tennessee area, lists prices on its treatment pages: traditional microneedling was US$375 a treatment there when checked in August 2026, with a course of three at $950. Disclosure: this site's publisher has an ownership interest in that business. No fee, commission or consideration is involved in the mention, and it is cited for one reason only — it prints a number where most of the industry prints "from". A reader should discount it accordingly, and should treat it as an order of magnitude rather than a quote: these are United States prices, and the figure in New Zealand, or in the next town, will be a different one.

The order of magnitude is the whole point. A course of a procedure costs what a shelf of serums costs, carries a practitioner, a consent form and a real complication rate, and is measured in trials of the kind set out above. That is the comparison worth making before anyone spends anything, in either direction.

How to use the ceiling

A serviceable test for any topical claim has three questions. What is the active, and roughly how big is it — because if it is a protein, a polysaccharide or anything described in kilodaltons, it is working on the surface or not at all. Where does it sit on the ingredient list relative to the preservative, because that bounds the concentration. And what was the outcome in the trial, measured how, over how long, against what control.

Applied honestly, that test leaves a shorter shelf and a clearer one. Sunscreen, daily, for prevention. A moisturiser chosen for its occlusive and humectant content rather than its hero extract. A small number of actives with trial evidence, used for months, accepting that the effects are modest and the irritation is real. Everything else is optional, and a good deal of it is pleasant, which is a legitimate reason to buy something — just not the reason printed on the box.