Every line on a product page is a public record, shown with the date we read it. Where there is no record, it says so. Where we have not been able to look, it says that instead — those are different things, and keeping them apart is the point of the whole site.
We do not score products. There is no number, no grade and no verdict anywhere on Rootify, because a score is our opinion wearing the clothes of a measurement. What there is instead: the record, who published it, and when we read it.
Three different kinds of evidence
A product page is short on purpose. These are the three things those short lines can be, and they do not carry the same weight.
An authority classification
Authority
A government or international body reviewed the published evidence on a substance and placed it in one of its own categories, or set a limit it considers safe. The body publishes the category; we copy it across with the date and a link, and translate the category into one plain sentence.
What it is not: It is not a finding about any product, and not a measurement of risk at the amounts in food. A category describes how strong a body judged the evidence to be. Two substances in the same category can carry wildly different real-world exposure.
For example: IARC places a substance in Group 2B. That means its working group found the evidence of cancer risk limited but not negligible — often from animal studies alone. It does not mean a product containing it will harm you.
A scientific study
Study
One piece of published research, read and summarised by a person, recorded with its citation, what it found, what kind of study it was, and who paid for it. Funding is recorded as its own field because a conclusion cannot be weighed without it.
What it is not: It is not a consensus. A single study is the weakest kind of evidence on this site precisely because it is the most specific — and where studies disagree, we record the disagreement rather than picking a winner.
For example: A 2019 paper reports an effect in rats at a stated dose. We say so, say who funded it, and link it. We do not conclude anything from it on your behalf.
A claim on the package
Label claim
Words printed on the packaging — "organic", "non-GMO", "no artificial colors" — as transcribed by a contributor to Open Food Facts. It tells you what the manufacturer chose to put on the box.
What it is not: It is not a certification and nobody audited it. Some of these claims sit behind a real third-party programme and some are unregulated marketing. Where we can tell the difference, we say which.
For example: A packet says "organic". That is a claim. The USDA Organic Integrity Database separately certifies operations — not products — and where we hold such a record we show it as its own, stronger line.
The authorities we read
Nine bodies, each with its own remit and its own categories. Knowing what kind of body published a classification is most of what tells you how to weigh it, so each one gets a sentence. Every entry links to where that body publishes its own assessments, so nothing here has to be taken on our word.
International Agency for Research on CancerIARC
The World Health Organization’s cancer research agency. Its working groups review the published evidence on a substance and place it in one of four categories. The category describes how strong the evidence is, not how dangerous the substance is at the amounts found in food.
California Office of Environmental Health Hazard AssessmentProp 65
California keeps a legal list of substances the state has determined can cause cancer or reproductive harm. A listing is a regulatory decision by one US state, and it triggers a warning label requirement there. It is not a finding about any particular product.
The European Union’s food safety body. It re-evaluates approved additives and publishes an opinion, often including a daily intake it considers safe. Its opinions are advice to EU regulators rather than law.
Joint FAO/WHO Expert Committee on Food AdditivesJECFA
A joint committee of the UN’s Food and Agriculture Organization and the World Health Organization. It is the international reference point for how much of an additive is considered safe to consume daily over a lifetime.
The US food regulator. Its inventory records which substances may be added to food and on what basis — including the ones permitted as “generally recognized as safe”, a status that does not require FDA review of every use.
Canada’s federal health department. It maintains its own lists of permitted food additives and the conditions on them, which differ in places from the US list.
French Agency for Food, Environmental and Occupational Health & SafetyANSES
France’s national food and health safety agency. It publishes its own opinions, which sometimes reach a different conclusion from EFSA on the same substance — a disagreement worth seeing rather than averaging away.
The EU’s chemicals agency. It holds the classification and labelling of chemical substances across all uses, not only food — so a hazard class here may describe industrial handling rather than eating.
French National Institute for Industrial Environment and RisksINERIS
A French public research institute on industrial and environmental risk. Its substance assessments are mostly about exposure outside food, so a record here is context rather than a food finding.
Every product page opens with a table of the same six questions. These are the answers it can give, and exactly how much each one claims.
Maker certifiedCertified organic
The company that makes this product appears in the USDA Organic Integrity Database as a certified organic operation, and we hold the certificate record and the date we read it.
It does not mean: That this item is certified organic. The register certifies operations — farms, handlers, processors — not individual retail products, and a certified handler can also make products that are not organic.
Not certifiedCertified organic
We searched the USDA Organic Integrity Database and found nothing for this product or its maker. US organic certification exists only by being in that register, so this is a definite answer: it is not certified organic.
It does not mean: That the food is not organically grown, or that anything is wrong with it. A small farm selling unsprayed produce without paying for certification gets this same answer.
Not verifiedNon-GMO verified
No verification programme we check lists this product. Verification is something a maker applies and pays for, so an absence is the complete answer to whether it has been verified.
It does not mean: That the product contains genetically modified ingredients. Most food has never been submitted for verification either way.
No recallsRecalls
No government notice in the records we hold — FDA, FSIS, CPSC and CBP — lists this product’s barcode.
It does not mean: That the product has never been recalled. We hold what those four agencies publish, going back as far as each one publishes, and a notice that never named a barcode cannot be matched to a product at all.
None flaggedOpen research
No ingredient on this label is one our classifier marks as an additive or processing ingredient.
It does not mean: That the ingredients are all safe, or that we have researched them. It is a statement about what kind of ingredients they are.
Not checkedAny question
We have not been able to look. Usually the brand has not been confirmed as a company yet, and every register we use is searched by company name.
It does not mean: Anything at all about the product. This is the one answer that is purely about us, and it is kept separate from “not certified” for exactly that reason.
Doesn't applyAny question
The question cannot sensibly be asked of this kind of product — organic certification of a baby bottle, for instance.
It does not mean: That we skipped it.
How we word a recall
A recall notice is written for regulators. “Product may be contaminated with Listeria monocytogenes” is precise and most people should not have to decode it. So a recall card leads with one plain sentence of ours — “May contain listeria, a germ that can cause serious illness” — and three rules govern it:
The agency’s own sentence is always still there. Our summary is shown beside it, never instead of it. You can check the paraphrase against the original without leaving the page, which is the only thing that makes paraphrasing defensible.
It restates, it never adds or softens. “May contain” stays “may contain”. We do not say a product is dangerous, do not say anyone was harmed, and do not reassure.
A notice we cannot summarise safely is printed as written. About one in ten. A borrowed summary that happens to be wrong is far worse than a technical sentence, so where the wording is unusual we leave it alone.
These summaries are written by a rule, not drafted by a language model, so every sentence the site can produce is fixed in advance and reviewable in one place rather than one notice at a time.
What we know we are missing
Measured against our own database on 7 October 2026. These are gaps we have found and not yet closed, listed because a filter that quietly under-reports is worse than no filter.
Our ingredient filters are only as good as our ingredient categories. Until 7 October 2026 several of the commonest additives in the catalogue had no category at all, so the filters silently missed them — a product containing soy lecithin passed a “no emulsifiers” filter. The rules now cover soy lecithin, xanthan and guar gum, locust bean gum, cellulose gum, caramel color, plain “vegetable oil”, “spices”, the sugar alcohols and the plant-derived sweeteners. The filters are named for what they now actually catch: “No added colors”, “No emulsifiers or gums”, “No sugar substitutes”. If you are filtering on something not named in a filter’s own description, read the ingredient list as well — the list is always the record, and the filter is a convenience over it.
Our study corpus is small. Twenty studies across sixteen ingredients. Authority classifications, described above, are how that widens — but an ingredient with no classification and no study has simply not been researched by us, which the ingredient page says in those words.
The organic register certifies operations, not products. A positive organic row means the maker holds a certificate, which is not the same as this item being certified. We say “maker certified” rather than “organic” for that reason.
What we will not do
Score a product. No grade, no number, no traffic light. Colours on this site say what kind of record exists, never whether something is good.
Treat an absence as a finding.“Not on file” means we looked and found no record. It is not a clean bill of health, and nothing on this site is ever green for an absence.
Blur a gap into a claim. If we have not checked something, the page says we have not checked, not that there is nothing there.
Draw the conclusion for you. Where authorities disagree — and they do — we show the disagreement rather than averaging it into a single answer.
If a record here is wrong, or has changed since we read it, we would rather hear it than not — see corrections. Every product also has its own page listing each source behind it with the date, reachable from the bottom of any product page.