How skin works
How long does skincare take to work?
Most actives show their first measured change between week four and week twelve; the humectants move the same day; the acids in days to weeks. Every number here is a trial’s first significant visit or its end, on one outcome, in a few dozen faces: a date the base can show you, not a promise for yours.
The short answer
The wait is a trial's date on one outcome, and how that date was found matters as much as the number.
- How long does retinol take to work?
- Give it four to eight weeks: that is when the change first became measurable. Measured on investigator-graded photoaging parameters versus vehicle; the first visit at which the change was measurable.
- How long does niacinamide take to work?
- Give it two to four weeks: that is when the change first became measurable. Measured on facial hyperpigmentation and skin lightness versus vehicle; the first visit at which the change was measurable.
- How long does vitamin C serum take to work?
- Give it eight to twelve weeks, and possibly less: that is simply when it was measured. Measured on investigator-graded photodamage on the treated half-face; measured at the trial's end, so possibly sooner.
- How long does hyaluronic acid take to work?
- This one works on contact: there is nothing to wait for. Measured on stratum corneum hydration; on contact, by mechanism.
The timeline: what each does, when
Weeks zero to twelve and beyond, the app’s three marks at day 7, 14, 30 as the ticks, and every bar the base’s own row: from the start of its range to its end, the outcome the trial measured written on it, and the bar’s end drawn by how the number was found. Pick any of the 158 ingredients the base has timed and it joins the four.
Give it four to eight weeks: that is when the change first became measurable.
how it was measured
Measured on investigator-graded photoaging parameters versus vehicle - lines and wrinkles, pigmentation, elasticity, firmness and overall photodamage on a 0-9 scale; the first visit at which the change was measurable. PMID 19852122PMID 17515510
Tucker-Samaras et al. tested a stabilised 0.1% retinol moisturiser once daily against its vehicle in a double-blind split-face study of women with moderate facial photodamage (36 on retinol, 28 on vehicle), grading at baseline, week 4 and week 8. At week 8 retinol was significantly better than vehicle on lines and wrinkles, pigmentation, elasticity, firmness and overall photodamage, and the paper records that many of those differences were already significant at the week-4 visit and improved progressively to week 8. The wait is therefore 4-8 weeks, with 8 weeks the ceiling for the full set of parameters; the paper does not say which individual parameters cleared significance at week 4, so the earlier figure is not claimed for wrinkles specifically. The 24-week 0.4% study usually quoted next to it is not a longer answer to the same question: it treated the inner upper arms of subjects with a mean age of 87 three times weekly and reported at 24 weeks. This is an industry study (Johnson & Johnson) with modest numbers, and 0.1% sits at the bottom of the 0.1-1% consumer range. Acne is a separate and slower wait, and the dryness and flaking most users notice in weeks 1-2 is irritation, not an outcome.
Give it two to four weeks: that is when the change first became measurable.
how it was measured
Measured on facial hyperpigmentation and skin lightness versus vehicle, computer image analysis and visual grading of high-resolution facial images; the first visit at which the change was measurable. PMID 12100180PMID 16029679
The vehicle-controlled clinical arms are in Hakozaki et al. 2002: 18 Japanese women with facial hyperpigmentation using 5% niacinamide moisturiser against its vehicle in a paired split-face design, plus 120 women in a facial-tanning study on vehicle, sunscreen, or 2% niacinamide with sunscreen. The paper reports that niacinamide significantly decreased hyperpigmentation and increased skin lightness compared with vehicle after 4 weeks of use, so 4 weeks is the earliest measured separation and the end of the wait, not the start of it. Improvement continues past that point: the 12-week 5% split-face study in 50 women takes facial images at 4-week intervals and adds fine lines, red blotchiness, sallowness and elasticity to the list, so ageing endpoints are a longer wait than pigment. Caveats are about confidence, not timing - the pigment panel was 18 people and single-ethnicity, both trials are Procter & Gamble work, and image-analysed spot area is not clinical melasma grading. Consumer products at 4-5% match the trial concentration, which is unusual.
Give it eight to twelve weeks, and possibly less: that is simply when it was measured.
how it was measured
Measured on investigator-graded photodamage on the treated half-face (fine wrinkling, roughness, tone); measured at the trial's end, so possibly sooner. PMID 10522500PMID 11896774
Two small randomised double-blind vehicle-controlled half-face trials carry this ingredient, and both report at 12 weeks. In the first, Cellex-C High-Potency Serum — ascorbic acid with acetyl tyrosine, zinc sulfate, sodium hyaluronate and green tea extract — ran against its own vehicle serum on opposite halves of 19 evaluable faces for three months, with optical profilometry of surface topography alongside clinical, photographic and patient grading; the contrast there is a multi-active serum against vehicle, not ascorbic acid isolated. In the second, 10% ascorbic acid plus 7% tetrahexyldecyl ascorbate in an anhydrous polysilicone gel ran against the same gel base in 10 patients, with clinical evaluation at weeks 4, 8 and 12; interim visits therefore existed, but separation was reported only at the endpoint (cheeks P=0.006, perioral P=0.01, overall P=0.01), so 12 weeks is a ceiling and true onset may be earlier. Both trials are very small. For pigment the picture is weaker rather than faster: 5% ascorbic acid against 4% hydroquinone in melasma over 16 weeks improved subjective scores but was clearly inferior to hydroquinone. Formulation matters more than duration here, since an oxidised or high-pH serum delivers nothing at any timepoint.
This one works on contact: there is nothing to wait for.
how it was measured
Measured on stratum corneum hydration, corneometer capacitance after a single application; on contact, by mechanism. Pavicic et al., J Drugs Dermatol 2011 (0.1% hyaluronic acid creams of different molecular weights, 8 weeks) for the wrinkle/elasticity arm. No named trial for the on-contact hydration rise; no PMID given.
Sodium hyaluronate forms a hygroscopic surface film that holds water against the skin, so the capacitance reading moves on contact; high molecular weight material does not cross the stratum corneum, so nothing deeper is being remodelled in that window. The other outcome the base hints at (anti_aging 1) is slower and weaker: topical 0.1% HA creams of differing molecular weight reported hydration, elasticity and wrinkle-depth improvement over an 8-week course. That trial used a specific low-MW fraction, not the generic HA in most serums.
- a solid end: the first visit at which the change was measurable: the number is an onset
- an open end: the trial's end: the number is a ceiling, and the change may have come sooner
- a dotted bar: the clinical consensus: no trial pins the point
- a dot at week 0: on contact, by mechanism: nothing to wait for
how it was measured
How long does retinol take to work
Four to eight weeks before investigator-graded photoaging parameters versus vehicle moved, and the number is the first visit at which the change was measurable. The trial is Tucker-Samaras et al., 2009, at 0.1%.
Give it four to eight weeks: that is when the change first became measurable.
What the trial measured, and in whom
- Tucker-Samaras et al. tested a stabilised 0.1% retinol moisturiser once daily against its vehicle in a double-blind split-face study of women with moderate facial photodamage (36 on retinol, 28 on vehicle), grading at baseline, week 4 and week 8.
- At week 8 retinol was significantly better than vehicle on lines and wrinkles, pigmentation, elasticity, firmness and overall photodamage, and the paper records that many of those differences were already significant at the week-4 visit and improved progressively to week 8.
- The wait is therefore 4-8 weeks, with 8 weeks the ceiling for the full set of parameters; the paper does not say which individual parameters cleared significance at week 4, so the earlier figure is not claimed for wrinkles specifically.
- The 24-week 0.4% study usually quoted next to it is not a longer answer to the same question: it treated the inner upper arms of subjects with a mean age of 87 three times weekly and reported at 24 weeks.
- This is an industry study (Johnson & Johnson) with modest numbers, and 0.1% sits at the bottom of the 0.1-1% consumer range.
- Acne is a separate and slower wait, and the dryness and flaking most users notice in weeks 1-2 is irritation, not an outcome.
How long does niacinamide take to work
Two to four weeks before facial hyperpigmentation and skin lightness versus vehicle moved, and the number is the first visit at which the change was measurable. The trial is Hakozaki et al., 2002, at 5%.
Give it two to four weeks: that is when the change first became measurable.
What the trial measured, and in whom
- The vehicle-controlled clinical arms are in Hakozaki et al. 2002: 18 Japanese women with facial hyperpigmentation using 5% niacinamide moisturiser against its vehicle in a paired split-face design, plus 120 women in a facial-tanning study on vehicle, sunscreen, or 2% niacinamide with sunscreen.
- The paper reports that niacinamide significantly decreased hyperpigmentation and increased skin lightness compared with vehicle after 4 weeks of use, so 4 weeks is the earliest measured separation and the end of the wait, not the start of it.
- Improvement continues past that point: the 12-week 5% split-face study in 50 women takes facial images at 4-week intervals and adds fine lines, red blotchiness, sallowness and elasticity to the list, so ageing endpoints are a longer wait than pigment.
- Caveats are about confidence, not timing - the pigment panel was 18 people and single-ethnicity, both trials are Procter & Gamble work, and image-analysed spot area is not clinical melasma grading.
- Consumer products at 4-5% match the trial concentration, which is unusual.
Niacinamide: the glossary page · how much niacinamide you need
How long does vitamin C serum take to work
Eight to twelve weeks before investigator-graded photodamage on the treated half-face moved, and the number is measured at the trial's end, so possibly sooner. The trial is Traikovich, 1999, at 10%.
Give it eight to twelve weeks, and possibly less: that is simply when it was measured.
What the trial measured, and in whom
- Two small randomised double-blind vehicle-controlled half-face trials carry this ingredient, and both report at 12 weeks.
- In the first, Cellex-C High-Potency Serum — ascorbic acid with acetyl tyrosine, zinc sulfate, sodium hyaluronate and green tea extract — ran against its own vehicle serum on opposite halves of 19 evaluable faces for three months, with optical profilometry of surface topography alongside clinical, photographic and patient grading; the contrast there is a multi-active serum against vehicle, not ascorbic acid isolated.
- In the second, 10% ascorbic acid plus 7% tetrahexyldecyl ascorbate in an anhydrous polysilicone gel ran against the same gel base in 10 patients, with clinical evaluation at weeks 4, 8 and 12; interim visits therefore existed, but separation was reported only at the endpoint (cheeks P=0.006, perioral P=0.01, overall P=0.01), so 12 weeks is a ceiling and true onset may be earlier.
- Both trials are very small.
- For pigment the picture is weaker rather than faster: 5% ascorbic acid against 4% hydroquinone in melasma over 16 weeks improved subjective scores but was clearly inferior to hydroquinone.
- Formulation matters more than duration here, since an oxidised or high-pH serum delivers nothing at any timepoint.
Vitamin C (LAA): the glossary page · how much vitamin c (laa) you need
How long does hyaluronic acid take to work
On contact: nothing to wait for; stratum corneum hydration moves at the first measurement, on contact, by mechanism. The trial is Pavicic et al., 2011, at 0.1% (no PubMed record named).
This one works on contact: there is nothing to wait for.
What the trial measured, and in whom
- Sodium hyaluronate forms a hygroscopic surface film that holds water against the skin, so the capacitance reading moves on contact; high molecular weight material does not cross the stratum corneum, so nothing deeper is being remodelled in that window.
- The other outcome the base hints at (anti_aging 1) is slower and weaker: topical 0.1% HA creams of differing molecular weight reported hydration, elasticity and wrinkle-depth improvement over an 8-week course.
- That trial used a specific low-MW fraction, not the generic HA in most serums.
Pavicic et al., J Drugs Dermatol 2011 (0.1% hyaluronic acid creams of different molecular weights, 8 weeks) for the wrinkle/elasticity arm. No named trial for the on-contact hydration rise; no PMID given.
Why the numbers differ: a first visit, a trial’s end, a consensus
The same “eight weeks” can mean two different things, and the base says which. Of the158 timed rows, 23 carry a number from the first visit at which the change was measurable: an onset, the honest kind. 32 carry the trial’s end: the study stopped there, so the number is a ceiling, and the change may have come sooner. 16carry the clinical consensus: no trial pins the point. And 87 work on contact, by mechanism: a humectant’s reading moves at the first measurement, and there is nothing to wait for.
- 23 first-visit rowsthe first visit at which the change was measurable: the number is an onset
- 32 trial-end rowsthe trial's end: the number is a ceiling, and the change may have come sooner
- 16 consensus rowsthe clinical consensus: no trial pins the point
- 87 on-contact rowson contact, by mechanism: nothing to wait for
Retinol’s four to eight weeks is the first kind; vitamin C’s eight to twelve weeks is the second, which is why its sentence says “and possibly less”. The bars above draw the difference at their ends.
What speeds nothing up
A higher dose is the first thing people reach for when the date feels far. Where a row’s basis speaks of dose at all, this is what it says; where it does not, the honest word is not measured.
- Retinol: The 24-week 0.4% study usually quoted next to it is not a longer answer to the same question: it treated the inner upper arms of subjects with a mean age of 87 three times weekly and reported at 24 weeks. This is an industry study (Johnson & Johnson) with modest numbers, and 0.1% sits at the bottom of the 0.1-1% consumer range.
- Niacinamide: Consumer products at 4-5% match the trial concentration, which is unusual.
- Vitamin C (LAA): In the second, 10% ascorbic acid plus 7% tetrahexyldecyl ascorbate in an anhydrous polysilicone gel ran against the same gel base in 10 patients, with clinical evaluation at weeks 4, 8 and 12; interim visits therefore existed, but separation was reported only at the endpoint (cheeks P=0.006, perioral P=0.01, overall P=0.01), so 12 weeks is a ceiling and true onset may be earlier.
- Hyaluronic acid: not measured. The row names no dose against its date.
- Benzoyl Peroxide: Irritation is front-loaded into those same two weeks, so the side effect usually arrives before the result, and 2.5% delivers the same lesion reduction as 10% with less of it.PMID 2948929PMID 26048131
- Betaine: The other half of its story is not: the osmoprotection and anti-irritant claims rest on in-vitro cell work, which the base's own evidence notes acknowledge, and no human study reports when an irritation or barrier endpoint changes with betaine at the 1-5% used in cosmetics.
- Pro-vitamin B5: Proksch & Nissen induced irritation with SLS in patch test chambers, then applied a 5% dexpanthenol cream or its vehicle twice daily and followed barrier repair, stratum corneum hydration, roughness and redness by serial biophysical readings: barrier repair was significantly faster than both vehicle and untreated skin, hydration higher, and redness reduced where the vehicle did nothing. Two limits on reading it as consumer reality: the population is healthy volunteers in an induced-irritation model, not patients with dermatitis, and 5% dexpanthenol is well above the 0.5-2% panthenol in most serums.PMID 19753737PMID 14641355
When to judge
Judge each row at its own end, not before: retinol at week 8, niacinamide at week 4, vitamin C at week 12, a humectant the day you apply it. The tool for that is a dated photo, not a mirror. My Shelf dates three marks from your first check-in - day 7, 14, 30 - asks for a photo again every 7 days, and speaks at day 7, 30, 90; the timeline above is the same clock, with the trials on it.
Sources
The four heads’ records, resolved on 2026-09-21; every other row’s sit in its fold on the timeline and on its own glossary page.
- Retinol, Tucker-Samaras et al., 2009: PMID 19852122, PMID 17515510
- Niacinamide, Hakozaki et al., 2002: PMID 12100180, PMID 16029679
- Vitamin C (LAA), Traikovich, 1999: PMID 10522500, PMID 11896774
- Hyaluronic acid, Pavicic et al., 2011: Pavicic et al., J Drugs Dermatol 2011 (0.1% hyaluronic acid creams of different molecular weights, 8 weeks) for the wrinkle/elasticity arm. No named trial for the on-contact hydration rise; no PMID given.
Corrections and updates
Nothing has been corrected on this page since it was published. If that changes, the change will be listed here with its date rather than quietly edited in.
Last re-read against its sources on . If you find an error, tell us and we will fix it and say so.
Get the next one
We read the primary papers, write to the researchers, and print what they say. If that is useful, the free plain-words routine guide comes with it. No spam, and nothing sold here.