Questions people actually ask

Think we have missed a brand?

Tell us. Every suggestion is read, and the ones that come up again and again go to the front of the queue — that is most of how the review list gets written.

Two things worth saying plainly. Suggesting a brand does not get it listed: it gets it read, and most of what we read does not make the shelf. And we do not take payment to review anything, so a brand writing in has exactly the same standing as a customer writing in.

Suggest a brand →

What we keep, and what we don't

The short version: we keep your email address if you give it to us, we keep nothing about your health, and the things you type into the Advisor are sent to answer your question and are not stored by us. Effective 18 September 2026.

Who we are

Aafiiya is operated by Kajol Patel, trading as Aafiiya, from Dubai, United Arab Emirates. For anything about your data, write to support@aafiiya.com. This policy covers aafiiya.com and both of its markets, .

What we collect, and why

An email address, if you give us one. There are three places you can: creating an account, signing up to the Letter, and asking for a reorder reminder. In each case we record the address, the market you were in, what you gave it for, and the time you gave it. If you create an account we also keep the name and username you chose. That is the whole record. Our sign-up code copies these fields out by name and discards everything else, so nothing further can reach our records by accident.

Nothing about your health. Your saved products, your goals, and any numbers you enter in Your Numbers are held in your own browser's storage and never sent to our servers. They are not attached to your account. Clearing your browser data removes them.

What you type into the Advisor. To answer a question the Advisor sends your conversation to Anthropic's API, which generates the reply. If you include a blood result or a symptom, it is part of that conversation. We do not store the conversation, log it, or attach it to your account or email address. Anthropic processes it under its own API terms in order to produce the answer. Do not put anything into the Advisor that you would not want a third-party service to process.

Your country, from the network. Our hosting provider tells our page which country a request came from, so the site can offer you the right market. We receive a two-letter country code and nothing else. It is not stored and is not tied to you.

Page views. We use Vercel Web Analytics to count visits. It does not use cookies and does not identify you.

Your password, honestly

If you set a password, it is kept only in your own browser, and it is kept unencrypted. It never reaches our servers. That is a weakness of the current build and we are fixing it. Until then, please do not reuse a password you use anywhere else.

Where it goes

Email records are held in Airtable. The site and its functions run on Vercel. The Advisor is answered by Anthropic. Fonts are loaded from Google Fonts, which means Google receives your IP address when the font is requested, as with any site that uses it. We do not sell your data, share it with brands, or use it for advertising. No brand pays to be listed, and no brand sees who signed up.

How long

Email records are kept until you ask us to remove them, or until you unsubscribe, whichever is first. Advisor conversations are not kept by us at all. Browser-side data lives as long as your browser keeps it.

Your rights

Under the UAE Personal Data Protection Law and India's Digital Personal Data Protection Act, you can ask what we hold about you, ask for it to be corrected or deleted, withdraw consent, and object to how it is used. Write to support@aafiiya.com and we will act within 30 days. If you are in India and are not satisfied with our response, you have the right to raise it with the Data Protection Board of India. If you are in the UAE, with the UAE Data Office.

Children

Aafiiya is for adults. We do not knowingly collect anything from anyone under 18, and the Advisor is not for use by or about children.

Changes

If this page changes, the date at the top changes with it. If the change affects what we collect, we will say so on this page rather than quietly editing it.

Terms of use

Plain-language terms for using aafiiya.com. Effective 18 September 2026. The privacy policy is a separate page and is part of these terms.

What Aafiiya is

Aafiiya scores supplements and wellness products against published evidence, the product label and the seller, and links you to places that sell them. We do not sell products ourselves. When you buy, you buy from the seller, on the seller's terms, and any problem with the order is between you and them.

Not medical advice

Nothing on this site, including anything the Advisor says, is medical advice, a diagnosis, or a substitute for a doctor or pharmacist. The scores are our reading of the evidence for a product's claims, not a statement that a product is right for you. Supplements can interact with medicines and conditions. If you are pregnant, breastfeeding, under medical care, or taking any medication, talk to a clinician before taking anything you read about here. Where the Advisor says a question needs a clinician, it means it.

How we make money

Some links to sellers carry an affiliate code, and if you buy through one we may earn a commission at no cost to you. Commission is agreed only after a product has been scored and can never change a score. No brand pays to be listed, to rank, or to be marked as selected. Prices shown are indicative and the seller's price at the time you buy is the one that applies.

Accounts

You can use the site without an account. If you create one, keep your details accurate and do not share your login. We can close an account that is used to abuse the service, the Advisor, or other people. You can ask us to delete your account at any time at support@aafiiya.com.

The Advisor

The Advisor is an AI system. It can be wrong. It answers from our catalogue and the sources we have opened, and it is instructed to say when the evidence is weak or missing, but you should treat its answers as a starting point for your own checking, not as a conclusion. Do not use it in an emergency. Do not use it to try to obtain anything it is designed not to give, and do not attempt to interfere with the service that runs it.

Our content

The scoring rubric, the product write-ups, the Letter and the design of the site belong to Aafiiya. You are welcome to quote from them with a link back. You may not copy the catalogue or the scores wholesale, or present them as your own. Brand names and product images belong to their owners and appear here to identify the products.

Accuracy and liability

We work from published sources and from labels as sold, and we correct mistakes when we find them or when you tell us. Even so, evidence changes, formulas change and labels change, and we cannot promise that everything on the site is current at the moment you read it. To the extent the law allows, Aafiiya is not liable for loss arising from your reliance on the site or on anything a seller does. Nothing here limits liability that cannot be limited by law.

Which law

For the UAE site, these terms are governed by the laws of the Emirate of Dubai and the federal laws of the United Arab Emirates. For the India site, by the laws of India. If a dispute cannot be settled by talking to us first, it goes to the courts of that jurisdiction.

Changes

If these terms change, the date at the top changes. Continuing to use the site after a change means accepting it. If you do not accept a change, stop using the site and, if you have an account, ask us to delete it.

Who we are and how we make money

Aafiiya scores supplements against the evidence, the label and the seller, and publishes the whole rubric so you can check our working. Nobody pays to be on the shelf.

How we make money

That is the whole arrangement. There is no paid placement, no sponsored ranking, and no brand has ever been told what it would need to change to score higher in exchange for money.

What we will not do

• We do not diagnose, treat or prescribe. Aafiiya is not a clinician. Where a question needs one, we say so and name a practitioner rather than selling you a capsule.
• We do not score what we cannot vouch for. A product with no authorised route to you gets no number at all — not a low one.
• We do not take a brand’s word for a label. Figures are read off the manufacturer’s own page. A reseller’s facts panel is a transcription by someone with no duty to keep it current.

When we get it wrong

We correct it in public and the score moves. In September 2026 we listed Thorne’s Vitamin D3 + K2 as NSF Certified for Sport. It is not — the brand runs its own testing programme with no outside certifier. We read their panel, corrected the entry, and the score went down.

Getting in touch

Corrections, challenges to a score, and brands who think we have read their label wrong: support@aafiiya.com. A challenge with a source attached gets looked at first.

The rubric, published in full.

A score that invites checking has to be checkable. So here is the whole thing — the gate, the four bars, what each one credits, and the decisions we have not settled yet.

What else we read

A score only means something next to what it beat. Every category is read end to end before anything goes on the shelf, and this is the whole ledger — listed against reviewed, category by category. The counts are the same ones printed on each product page.

What the card shows

Four scored bars, one gate above them, price stated underneath.

The Aafiiya score is the straight mean of the four bars, evenly weighted. There is no hidden weighting. If the displayed number does not match the mean, that is a bug, and the prototype checks every product against its own mean on load.

The gate: is there an authorised seller?

This runs before scoring and it is binary, not graded. An authorised route means the brand vouches for the chain: its own regional site, a distributor it names, or a retailer it acknowledges. Grey-market and parallel imports, marketplace third-party sellers, and anything untraceable to the manufacturer are not authorised routes.

Gate result	What happens
Pass	Scores normally. The line reads, for example, Available in the UAE · halal confirmed.
Caution	Scores normally with the caveat shown — an import, or marketplace listings only. Never merchandised.
Caution — international	No domestic seller, but the brand or a retailer it acknowledges ships here. Scores normally and keeps its buy link, with customs, duty and the longer wait stated. Never merchandised. A forwarder repacking stock is not this — that is a broken chain and fails on trust.
Fail — trust	No authorised route, or real counterfeit risk. Not scored, and no buy link. The card shows no number, a plain reason, and an explicit note that this is not a judgement on the product.
Fail — routing	On sale here, but not at a stockist we can link to — in the pharmacy, not on the site. Scores normally, no buy link, reason stated.

The two Fails are different animals. A high score sitting next to "you cannot buy this here" is an advertisement for the grey market: it spends our credibility on a purchase we refused to stand behind. So where the failure is about trust, there is no score. Where it is only about routing, there is nothing to protect anyone from, so the product scores normally and simply has no buy button.

Not scored means no number at all — not a zero. A 0.0 reads as "this product is terrible", which is the same misread the old 4.9 produced, only worse, and it would be a hand-placed figure that no bar produced: precisely the exception the gate exists to remove. The card says Not scored, with the reason underneath.

Why a gate and not a bar: a product you cannot safely buy is not a lower-quality product, it is a different category. Averaging that into a score is what forced the old hand-placed 4.9 cap. The gate removes the cap as a concept — nothing sits at 4.9 by rule any more. Products either score honestly or they do not score at all.

It is market-specific. The UAE question is mostly whether a distributor exists; the India question is counterfeit risk. The same brand can pass one and fail the other, so the reason always names the market. And it decays: distribution changes faster than formulation, so gate status carries its own recheck date. A route opening is good news worth surfacing.

Unscored products stay visible in search and in Advisor results. They are never merchandised. Showing them proves the work was done and captures the search; hiding them wastes both.

The four bars

Bar	What it measures
Is there evidence it works?	Scored per ingredient-claim, not per product, and inherited by every product carrying it. Vitamin D3 for deficiency correction is sourced once. Four published tiers, named on the card.
Is there enough in it?	Dose against the dose used in the supporting evidence. Within the studied range earns full credit; materially below earns little, however good the ingredient. A proprietary blend that hides the amount cannot be assessed, so it is scored as if absent.
Is it the right form?	Form against the form the evidence used. Magnesium oxide and magnesium glycinate are not interchangeable. Heat and humidity tolerance lives here too: a format that degrades in Gulf transit is the wrong kind of product to buy here.
Is it what the label says?	Every figure we publish comes from the company that makes the product. Ingredients, doses, forms and pack sizes are read off the manufacturer's own page — not a reseller's facts panel, which is a transcription by someone with no duty to keep it current and goes stale silently when a brand reformulates. A check that only ever saw a reseller is a partial check: the product still scores, but it cannot earn full credit here and it does not reach the shop window. Beyond the source, this bar is our verification work, not the brand's quality. Label checked against the manufacturer's own page, product confirmed in production, testing claims substantiated at source. An unsubstantiated claim earns no credit anywhere: "third-party tested" with no certificate we can open is treated as absent, not as a plus. Nothing is scored on the first three bars without at least a partial check on this one — grading a label nobody has read is fiction.

The evidence tiers

Graded, not strictly clinical. A strictly-clinical rule looks more rigorous but carries its own bias: it favours whatever is cheap, quick and ethical to randomise. It would mark down omega-3s despite an enormous observational base, and undercut the form distinctions Your Numbers is built on, since glycinate-versus-oxide evidence is largely absorption data. Grading is more honest about how nutrition research actually works — and it is more objective, not less, because every score has to declare its tier.

A bare number is not enough. "6.0" tells nobody anything, so the tier is named under the bar and links to the source behind it. Where the supporting evidence is exclusively industry-funded we show that separately, because it is a different problem from weak study design and deserves its own signal.

We grade the claim, not the vibe. "Supports immunity" and "prevents colds" are different claims and can earn different tiers from the same ingredient.

Price is stated, never scored

Price penalised premium products for being premium and rewarded cheap ones for being cheap, on a dimension that says nothing about whether a product works. It also moved with promotions, so the score wobbled every time a retailer discounted. Cost per day is live data, which is exactly why it had no business inside a number that implies stability.

So we show the price and the cost per day, and value moves to its own surface: same concern, same evidence tier, sorted by cost per day. Value becomes a way of navigating rather than a mark on a card, and the plain, cheap, correct product gets somewhere to win instead of losing a beauty contest it was never entered in.

The weakness line

One line under the headline score, same position, always present. Any bar below 8 names the weak bars — held back by dose and form. Nothing below 8 reads strong across the board. Exactly 8.0 passes; only below 8 is named.

It is a threshold rather than always-on because an always-on version would print "held back by form" on a product scoring 9.0 there, which is nonsense. And it never disappears: if it vanished on strong products, its presence would become the signal and people would read the absence instead of the content.

What earns a slot in the shop window

Concern pages, routines and the bestseller grid are merchandising. To appear there a product needs all of: an Aafiiya score of 8.0 or above; no per-product bar below 7 — dose, form and label; an evidence tier of 3 or better; bar four verified against the manufacturer rather than a reseller; and a gate Pass rather than a Caution. Then one more thing — it has to be genuinely the product we would recommend to someone we like. That last line is a veto. It may reject something that clears the first four. It may never promote something that fails them.

The floor of 7 covers the three per-product bars, not the evidence bar, and that distinction is deliberate. The floor exists to stop a good average hiding one broken leg — a quarter dose, the wrong form, a label nobody read. Those are facts about this product. The evidence bar is a fact about the science, inherited by every product carrying the ingredient, so a floor there would bar whole categories rather than individual products: tier 3 scores 6, and nobody is ever going to run a randomised trial proving you should replace the sodium you sweat out. So the evidence bar gets its own floor — tier 3 or better — which keeps tier 4 out of the window at any average.

Everything else lives in the scored catalogue: searchable, visible to the Advisor, not merchandised.

What the labels mean

These are the markers that appear on a product, and they are descriptions rather than scores — none of them moves the number.

Sources

Permitted: peer-reviewed journals indexed in PubMed; Cochrane reviews, systematic reviews and meta-analyses; and position statements from recognised bodies — NHS, NIH and its Office of Dietary Supplements, EFSA, WHO, national dietetic associations — as corroboration rather than as the primary basis.

Not permitted: brand-supplied evidence summaries and white papers, anything without a DOI or PubMed ID, press releases about studies, and secondary health-media coverage.

Verified means a named source with an identifier, opened and read, confirming the specific claim for the specific population at the specific dose. Not "a study exists". Institutional bodies summarise conservatively and slowly and mostly say nothing about specific forms or brands, so they corroborate; they do not carry a tier 1 score on their own.

The unfashionable-but-correct test

The rubric has to be able to say: this product is plain, cheap, badly photographed, and the right answer. Nothing here scores packaging, brand aesthetic, founder story or price. Price is displayed and never scored, and the best-value surface gives the plain correct answer somewhere to win.

Why India asks a different question

In the UAE the gate is mostly asking whether a distributor exists. In India it is asking whether what arrives is the thing on the label, and the answer changes with geography, because distribution is geography.

In the largest cities every brand is available and almost every claim is unverifiable; the ten-minute app delivers whatever ranks highest, which is a function of ad spend rather than evidence. In fast-growing smaller cities there are few authorised counters, so nearly everything is bought online — which is where counterfeits concentrate, and a tub arriving sealed tells you very little. In towns and rural areas the chemist is effectively the whole supply chain: the most trusted person in it, and also the person being sold whatever carries the best margin. Unbranded ayurvedic preparations are the specific concern there, for lead and mercury.

So the India shelf filters on the authorised seller, on whether that seller ships to the pincode, on a published FSSAI licence and on third-party heavy-metal testing. On the site itself this is one question — where are you — and a shorter list. The reasoning belongs here rather than on the homepage: someone arriving to buy magnesium has not asked for a lecture on supply-chain risk, and frightening them is not the same as informing them.

Two clocks

The label and the route decay at different speeds, so they are checked at different speeds. Running one clock for both is how a marketplace ends up sending someone to a dead link with a confident number next to it.

What	How often	Why
The label — bar four, the formulation, the testing claim	Every 6 months, read by hand off the manufacturer's own page	Formulations change slowly. The check date is printed on the product, not hidden — half the catalogue is always showing a check up to six months old, and saying so is worth more than pretending otherwise.
The route — the gate, the stockist, the buy link, stock	Every 14 days, automated; a person only looks at what breaks	Distribution moves fast. A dead link or a sold-out stockist is visible to a customer immediately, in a way a slightly stale label is not.

This sets the size of the catalogue, and we would rather say so than pretend the number is a strategy. A label read costs about half an hour and a route confirmation about ten minutes, so each product costs roughly an hour and twenty minutes a year to hold to this standard. Evidence sits outside that: it is scored per ingredient-claim and inherited, so it costs about ten hours a year however long the catalogue gets. That is the whole reason the rubric is built that way. The honest consequence is a catalogue in the dozens, not the thousands — and a product we cannot maintain to the standard does not belong here at all.

What we have not settled

Rubric v2. Supersedes v1, which invented A–D grades, scored price, scored evidence per product, and documented a 4.9 cap that the gate has now replaced.