How to build a skincare routine that actually works

Three products: a gentle cleanser, a moisturiser and a sunscreen. That is the core, and everything beyond it is optional. For cleansing, NICE’s acne guideline (NG198) recommends a non-alkaline syndet cleanser — a synthetic detergent formulation, pH-neutral or slightly acidic — used twice a day. The advice is written for acne-prone skin, but it is a sensible default for anyone, because ordinary soap is alkaline and skin is not. The same guideline advises acne-prone skin to avoid oil-based or comedogenic moisturisers, sunscreens and make-up, and to remove make-up at the end of the day.

The sunscreen rules come from the NHS, and they are specific: at least SPF30, with at least 4-star UVA protection or the UVA circle logo on the label. Quantity matters more than brand. The NHS suggests six to eight teaspoons for the whole body, applied 30 minutes before going out and again just before you leave, then reapplied every two hours and after being in water. For the face alone, the American Academy of Dermatology suggests roughly a teaspoon.

Notice what those recommendations are about: what a product is (non-alkaline, non-comedogenic, SPF30 with strong UVA cover), never what it costs. Formulation is the selection criterion. Where price does and does not buy real performance has actual laboratory data behind it, and we cover it separately in what’s worth paying for in skincare.

The American Academy of Dermatology’s published order is short: cleanse first, then any active or treatment — in its words, “apply medication or treatment right after cleansing” — then moisturiser and/or sunscreen.

Here is the honest note you will rarely read: no randomised trial has ever compared layering orders. “Thin to thick” and its variants are practical consensus, not tested science. Only two ordering rules have a real mechanistic basis. First, retinoids belong at night. Second, sunscreen goes on last every morning, because it has to sit at the surface to do its job. Beyond those two, the order you will actually stick to is the right one.

Once the core routine is in place, a short list of actives has respectable evidence. One theme recurs: the trials were run at modest concentrations, so the percentage arms race on ingredient labels has less science behind it than the packaging implies.

For dark marks and uneven tone: niacinamide or azelaic acid

Niacinamide has two of the better-known trials in cosmetics. A 5% preparation reduced hyperpigmentation from around four weeks (Hakozaki and colleagues, British Journal of Dermatology, 2002), and a 12-week trial in 50 women found improvements in fine lines, spots, texture and blotchiness at the same 5% strength (Bissett and colleagues, 2004–2005). Two caveats belong next to those results: the trials were small and largely funded by Procter & Gamble, whose formulations were being tested, and the evidence sits at 2–5% — there is no trial behind the 10%-plus products. Azelaic acid stands on firmer ground for acne and rosacea: a 2023 systematic review of 43 randomised trials found it beat vehicle (the same product minus the active) for both conditions at 12 weeks (King and colleagues, Journal of Cosmetic Dermatology). In the UK, the 15% and 20% strengths are prescription products; around 10% is sold in cosmetics.

For sun damage: vitamin C, worn under sunscreen

The review evidence (Al-Niaimi and Chiang, Journal of Clinical and Aesthetic Dermatology, 2017) puts the useful range for L-ascorbic acid at 8–20%, with absorption plateauing rather than improving above that. In one study, a 10% preparation roughly halved UVB-induced skin reddening (a 52% reduction), and vitamin C performs measurably better combined with vitamin E and ferulic acid. Its role is precise: an antioxidant layer under sunscreen, never a substitute for it.

For texture: exfoliating acids, with modest expectations

Alpha hydroxy acids do something real but bounded. An 8% glycolic or lactic acid cream produced modest, measurable improvement in photodamaged skin in a randomised trial (Stiller and colleagues, 1996). But when a 2025 network meta-analysis in Scientific Reports pooled the available comparisons, glycolic acid was not significantly better for fine wrinkles. Treat AHAs as texture-and-tone tools, clearly weaker than retinoids for wrinkles. Salicylic acid (BHA) has modest, lower-grade evidence for mild acne and is not in NICE’s recommended first-line options.

For dryness: ceramides and hyaluronic acid, for comfort not cure

In a randomised trial in moderate eczema, a ceramide moisturiser improved water-loss and hydration measurements against placebo over 28 days — but eczema severity scores were not significantly different by day 28 (Spada and colleagues, 2021). That is the honest reading: real barrier support, not a treatment. Hyaluronic acid is a humectant: it draws water into the surface of the skin and plumps it temporarily. Useful and pleasant, but it is hydration, not structural change.

The best-evidenced active family of all, the retinoids, comes with enough UK-specific complication — what is legal to buy, what needs a prescription, what the EU rule really says — that it has its own guide.

The most useful number in skincare is twelve weeks. When NICE sets out first-line acne treatment — prescription-strength combinations — it specifies 12-week courses with a review at the end. If prescription medicines are given three months before anyone judges them, a cosmetic cream deserves at least the same patience. Most products are abandoned long before they have had a fair trial.

The trial timescales above tell the same story. Niacinamide’s pigmentation effect appeared from around four weeks, with line and texture results measured at twelve. Azelaic acid’s review evidence sits at twelve weeks. Tretinoin’s photoageing trials typically ran sixteen to twenty-four weeks. Only hydration moves quickly: the ceramide trial’s moisture measurements shifted within 28 days. As a rule of thumb: hydration moves first, breakouts need about three months, and fine lines need four to six.

Fewer than the internet says. The one combination with real chemistry behind it involves benzoyl peroxide and tretinoin: applied at the same time, benzoyl peroxide oxidises old-style tretinoin, degrading around 50% of it within two hours and around 95% over 24 hours in light (Martin and colleagues, British Journal of Dermatology, 1998). But the detail matters. Adapalene — the retinoid most often paired with benzoyl peroxide — showed no degradation over 24 hours in the same work, and the two are sold together as a licensed fixed combination. Modern stabilised tretinoin formulations also resist the effect. The practical rule: with an older tretinoin preparation, keep benzoyl peroxide to the morning and tretinoin to the night; otherwise this famous “clash” mostly no longer applies.

The genuine risk when stacking actives is not chemistry but arithmetic: irritation adds up. There is no good trial of retinoid-plus-exfoliating-acid regimens, and NICE’s approach to retinoid irritation is the sensible template for any strong active — start at low strength and low frequency, and build up as your skin tolerates it.

One famous incompatibility can be retired altogether. The claim that niacinamide and vitamin C cancel each other out traces back to 1960s experiments that combined niacin and ascorbic acid in solution at high heat — conditions with no relevance to skin, as a 2014 review pointed out (Wohlrab and Kreft). Similarly, “vitamin C only works in the morning” is false: when you apply it is a matter of preference, provided that in the daytime it sits under sunscreen rather than standing in for it.

“Steam opens your pores.” Pores have no muscles, so they cannot open or close — whatever steam or cold water may do for your skin, it is not that.

“Detox” skincare. Sense About Science examined detox products across industries and put it flatly in its Detox Dossier: “‘detox’, as used in product marketing, is a myth”.

“Chemical-free.” This one is citably meaningless. In 2010 the Royal Society of Chemistry offered £1 million to anyone who could produce a genuinely chemical-free product. The bounty was never claimed. As the RSC’s Dr Richard Pike put it: “everything we eat, drink, drive, play with and live in is made of chemicals”.

A useful habit follows from all of this: when a label leans on words like these, or on a headline percentage, look for the trial instead — and give whatever you choose its twelve weeks.