What the evidence said, and where it changed my mind.

Read, not
assumed.

All accessed 3 August 2026. Where a source was weak I made the on-site claim weaker to match, rather than dressing it up. That is most of the method.

What I could not reach

• Trustpilot and Boots.com both returned 403 to automated fetching, so verbatim UK retail review sentences are thinner here than I wanted. Source 1 gives tag prevalence across 324 reviews but not the sentences people wrote.
• Reddit blocked fetching, which removes the richest source of the exact words customers use.
• PubMed Central served a CAPTCHA, so the niacinamide concentration literature is represented second-hand through source 13. The site's copy is worded to survive that: it says the data plateaus around 4 to 5% rather than citing a specific trial.
• Mintel's UK facial skincare reports are paywalled. UK-specific category spend is therefore missing, and source 9 is US data used directionally rather than as a UK figure.

1. 01

   The Ordinary Niacinamide 10% + Zinc 1%, aggregated user reviews

   SkinSort

   324 reviews, 3.1 / 5. Reviewer-applied tags with prevalence: Great Value 32%, Irritating 29%, Broke Me Out 27%.

   https://skinsort.com/products/the-ordinary/niacinamide-10-zinc-1
2. 02

   Niacinamide 10% + Zinc 1% product page, UK

   The Ordinary (Deciem)

   £5.00 for 30 ml. Percentages sit in the product name; the ingredients block does not restate them. Claims visible results in 7 days.

   https://theordinary.com/en-gb/niacinamide-10-zinc-1-serum-100436.html
3. 03

   Moisturising Gel Cream product page

   BYOMA

   $14.99 / 50 ml, 4.55 from 549 reviews. An INCI Decoded panel groups ingredients by function and publishes no concentrations.

   https://www.byoma.com/products/moisturizing-gel-cream
4. 04

   Skin Barrier (Dr Melissa Piliang, MD), 14 December 2022

   Cleveland Clinic

   Causes of damage listed as harsh soaps, over-exfoliating or scrubbing, and not moisturising. Recovery described as three to four months.

   https://health.clevelandclinic.org/skin-barrier
5. 05

   5 Expert Tips for Exfoliating Without Disrupting the Skin Barrier

   CeraVe

   Warns that over-exfoliating causes redness, flaking and irritation, and then gives no frequency guidance at all.

   https://www.cerave.com/skin-smarts/skincare-tips-advice/5-expert-tips-for-exfoliating-without-disrupting-the-skin-barrier
6. 06

   Cart abandonment: documented reasons

   Baymard Institute

   Extra costs too high (shipping, tax, fees) 40%, delivery too slow 20%, no trust in card handling 19%, forced account 18%, cannot see total upfront 12%.

   https://baymard.com/lists/cart-abandonment-rate
7. 07

   Sunscreen and sun safety

   NHS

   UK consumers are taught the 5-star UVA system and household quantity measures, not ISO test values.

   https://www.nhs.uk/live-well/seasonal-health/sunscreen-and-sun-safety/
8. 08

   Reference pricing, published 12 December 2016

   ASA / CAP

   Comparison prices must be genuine and previously charged for a reasonable period, with the material information a shopper needs to understand the comparison.

   https://www.asa.org.uk/news/reference-pricing.html
9. 09

   Beauty demand holds, but price concerns reshape spending, 29 August 2025

   eMarketer, citing Circana

   Only 14% of beauty buyers think a higher price means better quality. Mass beauty grew 4% in H1 2025 against 2% for prestige.

   https://www.emarketer.com/content/beauty-demand-holds-price-concerns-reshape-spending
10. 10

    68% of consumers seek clean beauty information

    Cosmetics Business, reporting NPD Group

    60% expect brands to disclose ingredient sources; 72% want clear explanations of what ingredients do. No sample size or field date is published, which is a real weakness.

    https://cosmeticsbusiness.com/68-of-consumers-seek-clean-beauty-information-and
11. 11

    Retinol vs Retinal

    The INKEY List

    Retinal converts to retinoic acid in one step where retinol takes two; the brand claims results up to 11x faster than retinol.

    https://www.theinkeylist.com/blogs/news/retinol-vs-retinal-what-s-the-difference-and-which-one-should-you-use
12. 12

    Reader comments on product pilling, 29 April to 12 May 2025

    Skincarma

    A reader posting as H2theD reports product balling happening daily and asks where the tipping point is; the host, Carmine Montalto, names mineral sunscreen as one possible culprit.

    https://skincarma.substack.com/p/the-skincarma-pod-episode-23-cracking/comments
13. 13

    Niacinamide, ingredient reference

    SkinSort

    Around 5% is cited for pigment benefit; true niacin flush does not occur with niacinamide; market products run from 0% to 20%.

    https://skinsort.com/ingredients/niacinamide
14. 14

    Are Beauty Brands Intentionally Hiding Their Ingredients?, 25 June 2020

    Jessica DeFino

    Disclosure law predates e-commerce, so 'transparency' has become a positioning claim that costs a brand nothing to make.

    https://jessicadefino.substack.com/p/are-beauty-brands-intentionally-hiding-their-ingredients-694b38f0f11d

Method,
and why
that method.

Sixty minutes,
nothing leading.

The craft is in what is not asked. No question here supplies an answer, offers examples to pick from, or asks anyone to predict their own future behaviour. If a participant never mentions percentages, that silence is the finding and the guide does not rescue it.

1. 1. Warm-up

   5 min

   Get them talking about their own routine before the topic narrows.

   • Talk me through what you put on your face this morning, in order.
   • How long has that been the routine?
   • What changed it last?

   Probe on the last item only: what happened, when, and what they did next. Do not ask whether it was a good change.

2. 2. Context

   12 min

   Understand where products come from and who they trust, without naming any brand first.

   • Where did the products you use now come from? Take them one at a time.
   • Tell me about the last time you bought something for your skin that you had not used before.
   • Who or what do you check before you buy something new, if anything?
   • When did you last stop using something? What made you stop?

   If a source is named, ask what makes that source useful rather than whether they trust it. Never supply examples: no 'do you use Reddit or TikTok?'

3. 3. Deep dive

   20 min

   Get at the mental model around ingredients, concentration and irritation without teaching it.

   • Show me the back of something you are using. Read me anything that catches your eye.
   • What does that tell you, if anything?
   • You mentioned [ingredient]. How did you come to know about it?
   • Tell me about a time a product did not agree with your skin. What did you do?
   • How do you decide how much of something to use, or how often?
   • Last time you compared two products, what were you comparing?

   When a percentage comes up: 'what would a different number there mean to you?' Not 'is higher better?', which supplies the answer. If they do not raise percentages at all, that is a finding; do not rescue it.

4. 4. Reaction

   15 min

   Reaction to the artefact, elicited by task rather than by opinion.

   • Here is a product page. Spend as long as you like and tell me what you make of it.
   • Find out how much niacinamide is in it.
   • You are buying this one item and nothing else. What will it cost you delivered?
   • What would you want to know that is not here?
   • Who do you think this is for?

   Silence is the tool. Where they stall, note the elapsed time and the element they were looking at, then ask 'what were you expecting there?' Never ask 'do you like it?' or 'is that clear?'

5. 5. Wrap-up

   8 min

   Catch what the structure missed.

   • If this brand existed, what would have to be true for you to buy it?
   • What would stop you?
   • Is there anything I should have asked and did not?

   Ask the last question and then wait. It routinely produces the most useful thirty seconds of the session.

Six themes,
counted.

Prevalence is expressed as a count of sources, not as a percentage of people. Fourteen sources is a small denominator and a theme carried by two of them is labelled as thin, because it is.

Who this is for

• The list-reader

  primary, and the one the page is built for

  Reads INCI lists, knows what phenoxyethanol is, has an opinion about fragrance. Evidenced by the existence and traffic of ingredient-decoding tools, and by competitors shipping decoded-INCI panels to satisfy exactly this person. Buys on specifics and is repelled by atmosphere.

• The barrier-repair arrival

  secondary

  Arrives after something went wrong, which the clinical sources describe as a three-to-four-month recovery. Wants to know what to stop doing more than what to start. This is why the thesis section argues subtraction before the range is shown at all.

• The value-anchored comparer

  the one most likely to bounce

  Has a £5 serum in the cupboard that 32% of its reviewers call great value. Everything on the page has to survive being read next to that. Named here as a hypothesis: it is inferred from review-tag prevalence and category growth data, not from anyone I have spoken to.

1. T1surfaced in 5 of 14 sources

   Irritation, not inefficacy, is the category's failure mode

   The most-reviewed niacinamide serum in the category holds 3.1 out of 5 across 324 reviews, and the two commonest complaint tags are Irritating (29%) and Broke Me Out (27%). Cleveland Clinic puts harsh soaps and over-exfoliating at the top of what damages a barrier. CeraVe warns about redness and flaking from over-exfoliation. Nobody in the set complains that a serum did nothing; they complain it did something unwanted.

   over-exfoliating or exfoliating incorrectly can disrupt the skin barrier and result in redness, flaking, or irritation

   ImplicationThe competitive opening is a lower dose stated confidently, not a higher one. A brand that says 4% and explains why is arguing against the thing 29% of reviewers actually experienced.

2. T2surfaced in 2 of 14 sources, and it is the thinnest-sourced theme here

   Texture under sunscreen is a named, specific complaint

   Pilling, or balling, comes up as a lived daily problem in a public comment thread among people who follow formulation closely, with mineral sunscreen named as a likely culprit. It is a texture failure that only shows up in the third layer of a routine, which is exactly where a product page never tests itself.

   a reader posting as H2theD reports product balling happening daily and asks where the tipping point is; the host names mineral sunscreen as one possible culprit

   ImplicationWorth saying out loud on the page, but worth flagging as the finding I would most want to test with real people before leaning any harder on it. Two sources is a hypothesis, not evidence.

3. T3surfaced in 4 of 14 sources

   Transparency has become a claim rather than a practice

   The category's cheapest serum puts its percentages in the product name and then does not restate them anywhere in the ingredients block. A barrier-led competitor ships an INCI Decoded panel that groups ingredients by function and publishes no concentrations at all. Reported survey data has 72% wanting clear explanations of what ingredients do, against 60% wanting disclosure of where they come from. The appetite is for the explanation, and the explanation is the part nobody ships.

   72% want clear explanations of what ingredients do

   ImplicationPublishing the full w/w sheet is only half a differentiator. The other half is a sentence per row saying why that number and not another one.

4. T4surfaced in 3 of 14 sources

   Price has stopped signalling quality on its own

   Only 14% of beauty buyers think a higher price means a better product. Mass beauty grew twice as fast as prestige in the first half of 2025. And the top tag on a £5 serum with 324 reviews is Great Value, at 32%, which is higher than either complaint tag. A £48 serum cannot rely on looking expensive; the £5 one has already claimed the value ground and has the reviews to hold it.

   Only 14% of US beauty buyers think that higher prices mean better quality products

   ImplicationThe premium has to be itemised. Per-ml price, batch size, fill location and what the money buys, stated as figures rather than implied by whitespace.

5. T5surfaced in 1 of 14 sources, but it is the only quantitative behavioural study in the set

   The basket is lost on cost, not on brand

   Extra costs at checkout are the single largest documented reason people abandon a basket, at 40%, ahead of trust in card handling, forced account creation and checkout length. A further 12% abandon because they cannot work out the total cost upfront.

   Extra costs too high (shipping, tax, fees)

   ImplicationDelivery terms belong above the brand argument, not in the footer, and the shortfall to free delivery should be a live number rather than a banner.

6. T6surfaced in 2 of 14 sources, by contrast rather than agreement

   Nobody will tell you how often to use an active

   A mass brand's own guide to exfoliating without wrecking your barrier warns about the consequences and then declines to name a frequency. The NHS, with nothing to sell, gives quantities in teaspoons and a two-hourly reapplication interval. The commercial incentive not to say a number is visible in the gap between those two documents.

   As a guide, adults should aim to apply around 6 to 8 teaspoons of sunscreen if you're covering your entire body.

   ImplicationPrinting an actual schedule, two nights a week and which nights, is a differentiator that costs repeat sales and is therefore credible.

Insight to opportunity

Six things
I had to
go back and fix.

1. 01

   I removed a strikethrough price, and that was the sharpest thing this research did

   The routine builder showed the full AM set at £112 struck through beside £74, because 06 is out of stock. CAP guidance is that a comparison price must be genuine and must have been charged for a reasonable period. £112 was never charged and never withheld; nothing is on offer. On a site whose entire argument is that it does not mislead you, that was the worst thing on the page. It now reads 'Full set £112 once 06 is back on 19 August', in plain type.

2. 02

   Delivery moved above the brand argument

   The hairline strip carrying free-delivery threshold, lead time, dispatch location and returns window sits directly under the hero, ahead of any argument about why the range is small. That ordering is a direct read of the abandonment data: extra costs beat trust, account creation and checkout length as a reason to leave. The cart drawer got a live shortfall figure for the same reason, and the hero says out loud that £48 is £2 under the £50 line rather than hoping nobody notices.

3. 03

   Dead ends became live buttons

   Out-of-stock 06 rendered a greyed 'Unavailable' in the list and a disabled button labelled 'Email me when back' in the quick view, which promised an action the code refused to perform. In a category losing four in ten baskets to friction, a dead control on the one product that is genuinely unavailable is indefensible. Both are now live: they scroll the restock form into view and move focus into the field.

4. 04

   The niacinamide argument stopped being an assertion

   The formulation note used to say 4% not 10% because the data plateaus. It now also carries the review-tag prevalence from the category leader: 29% of 324 reviewers tag it irritating. Same conclusion, but now it is checkable by anyone who clicks source 1.

5. 05

   A finding contradicted me, and the sunscreen copy changed

   I had 06 Mineral Screen leading on UVA-PF 11.4 under ISO 24443, which is the number a formulator cares about. The NHS teaches UK shoppers a five-star system and gives quantities in teaspoons, so the ISO figure lands on almost nobody. The blurb now gives the lab value and the star equivalent, and the AM routine step measures the dose in fingers rather than mg/cm². The research said my instinct for specificity was pointing at the wrong specific.

6. 06

   I kept a claim smaller than the competition's

   A UK competitor's own copy says retinal works up to 11x faster than retinol. 05 Retinal claims that 0.1% does roughly what a 0.5% retinol does, which is a fifth of that multiple. Where the category inflates, the credible move for this brand is to under-claim and let the reader notice.

Nine colours,
one radius.

All of it is declared in one `@theme` block in globals.css. There are no ad-hoc hex values anywhere in the components.

Colour

• --color-bone#F4F1EAPage ground
• --color-bone-2#ECE7DCFooter, announcement bar, row hover
• --color-ink#16150FBody type, primary buttons, Oslo band
• --color-stone#6B6455Muted copy, interactive borders. 5.20:1 on bone
• --color-olive#4A5240The only accent. Labels, bars, focus ring. 7.24:1
• --color-olive-2#3A4132Hairlines inside the ink band
• --color-clay#C4B7A2Hairlines; body copy on ink at 9.27:1. Never type on bone
• --color-clay-2#DDD5C5Row rules and bar troughs. Decoration only
• --color-rust#8A4B2FLow stock and form errors. 5.95:1 on bone

Type

• DisplayInstrument Serifclamp(2.05rem, 7.2vw, 5.2rem), leading 0.84 to 0.92
• BodyInter 40013 to 15px, leading 1.75 to 1.85
• Micro-labelInter 50010px, 0.16em tracking, uppercase
• NumeralsInter, tabularfont-variant-numeric: tabular-nums on every figure

Spacing, radius, motion, focus, targets

• SpacingTailwind 4px scaleSections py-16 at base, py-24 to py-28 from lg
• Containermax-w-[1500px]px-5 / sm:px-8, plus lg:pl-[88px] and xl:pl-[110px] for the rail
• Radius2pxButtons only. Nothing else on the site is rounded
• Motion400ms cubic-bezier(0.16, 1, 0.3, 1)Every transition. Disabled under prefers-reduced-motion
• Focus2px olive, 3px offsetClay inside .on-ink, where olive would be 2.24:1
• Targetmin-h-11 (44px)Every interactive element

Component: Button