How ROOTED rates things.

This page has two jobs. It tells you how a rating is reached, so a tier is something you can check rather than a number you have to trust. It is the page our Terms of Use point to: ROOTED publishes opinions about named products and businesses, and those opinions rest on the criteria written here.

The tiers

Every rated product sits at one of three tiers. A tier is our reading of one product, as it was sold on the day we checked it, against the published standard for its shelf. It is not a score, and it does not attach to the brand or to the shop that sells it.

Two different things can sit against a product, and we keep them apart, because they mean different things to you.

A compromise is something about the product. A synthetic fibre in the blend. A performance finish. A plastic layer inside a cotton protector. It is a property of the thing you would receive, and it is a real tradeoff you might accept or reject.

A gap is something about the page. The brand did not name the factory, or did not say where the garment was made, or did not say what the dye was. It is not a fault in the product. It is a question we could not answer, and we would rather write that down than guess.

Most gaps are just gaps. A small maker who sews in one workshop and never wrote it down is not hiding anything. But a gap is also the reason we sometimes cannot go further than we do, and you are entitled to know which of the two you are looking at.

Tier 1, Rooted Approved. The product meets the published standard for its shelf and carries no compromise we consider meaningful. Anything the brand left undisclosed is listed on the entry as a gap, so a Tier 1 product can still have blanks on its page. It means we found nothing wrong, not that we know everything. Certifications shown are the ones the brand holds; none is required for Tier 1.

Tier 2, Good Option. The product meets most of the standard and carries one or two compromises. Each one is named on the entry in plain words. If you open a Tier 2 product and cannot find the compromise, that is a mistake on our side and we want to hear about it.

Tier 3, Budget/Transitional. The product carries one specific compromise that we name. It is here either because it is the mainstream default with its tradeoff made visible, or because it is a cheaper option that is still decent. Tier 3 is not a bargain bin, and some Tier 3 products cost more than Tier 1 products on the same shelf.

What a tier is not. It is not a safety rating and it is not a health claim.

Some categories cap a product by rule. A cap is different from a gap: a gap is a fact the brand did not publish, and a gap never lowers a tier; a cap is a property of the product the standard does not allow at the higher rung, and the cap governs. A named component whose composition is not disclosed looks like a gap and is a cap in effect, because the standard cannot be applied to a component it cannot see. The 95 percent natural-fiber threshold works the same way: a main fabric at 95 to 99 percent natural meets the material threshold, and naming the remaining synthetic as a compromise on the entry caps it at Tier 2. Those are two different facts and both are true: a gap never lowers a tier, a cap is not a gap, and the cap governs. Where a category keeps a testing requirement, the distinction is this: a certification requirement that only proves somebody filed paperwork is an empty gate and comes out; a requirement that is the only evidence about a specific hazard stays. Greens powders keep third-party heavy-metal testing and nonstick cookware keeps third-party testing for that reason. Fluoride toothpaste is capped at Tier 3 no matter how good the rest of the formula is. A product containing anything on our hard floor is not listed at any tier: PFAS, parabens, phthalates, formaldehyde donors, triclosan, oxybenzone or octinoxate, undisclosed fragrance, aluminum antiperspirant, seed oils in food or in animal feed, and wrinkle-resistant chemical finishes.

A tier attaches to one product as sold on the date we checked it. It does not attach to the brand, to the brand's other products, or to any store that sells it. A store that stocks Rooted Approved products is not itself Rooted Approved.

Where we are behind. Many entries on the site still carry a general description of the material rather than a reason for the tier. We are rewriting those. And although a compromise and a gap mean different things, an entry today lists them together in one list of cons, so the difference this page draws is not yet one you can see on the product. We are separating them. Until an entry says why it sits where it sits, take the tier as our judgment rather than as an argument you can check, and read the pros and cons underneath it, which are what we actually found.

How a product is assessed

Each of the 172 subcategories on the site has its own published standard: what we are looking for, green flags, red flags, and what to avoid. A product is rated against its subcategory standard, not against a general impression. The standard is on the subcategory page.

An entry records what we found: a verdict, the composition as published by the brand, flags, pros, cons, and the certifications the product holds. Where a price is an estimate, the entry says so.

What counts as evidence, and what does not. A claim needs evidence that could actually establish it. Our checks are specific:

• An SPF number is a measured result. The manufacturer's statement that it tested the product is recorded as the manufacturer's statement.
• A whitening or remineralization claim needs a study. An ingredient list is not one.
• A germ-kill claim needs EPA registration or efficacy data. A named active ingredient is not efficacy.
• A contaminant-removal claim needs certification or testing for that contaminant. NSF/ANSI 42 covers taste and odor only and never substantiates a health claim.
• A flicker-free claim needs an IEEE 1789 measurement. A blue-light percentage needs wavelength data. An EMF claim needs a measurement.
• A heavy-metals or residue claim needs a certificate of analysis or independent lab testing.
• "Organic" needs a named certifying body. "Certified organic" with no body named is treated as the brand's wording, and the entry says so.
• "Non-toxic", "chemical-free" and "toxin-free" are never accepted. They are marketing language that hides the formulation. We describe the formulation instead.
• "Detox" and "hypoallergenic" are never accepted, for the same reason.
• Certification in progress is an intention, not evidence.
• EWG Verified, B Corp, Leaping Bunny, Fair Trade, FSC and NSF Certified for Sport are real credentials that say nothing about whether a product does what it claims. They are shown for what they are.

When a brand makes a claim we cannot verify, we do one of three things. We cite evidence if it exists, with the source and date. We record the claim as the manufacturer's statement if the claim is regulated and the responsibility is theirs. Otherwise we remove the claim and describe the product instead. We do not soften a claim into a vaguer version of itself.

How this is enforced. A check runs before every publish. Any Tier 1 product carrying a claim its recorded evidence does not support blocks the publish until the claim is fixed, cited, or reviewed with a written reason and date. On Tier 2 and Tier 3 products the same finding is a warning rather than a block; we cleared every one of those by hand on 2026-09-15, and this page will say so if that changes.

A tier is an approved act, not a default. A product can carry a tier only if its id is in a dated approvals ledger in the site's data; a push that gives a product a tier without that entry fails. A product whose rating has been drafted but not yet approved shows "Rating pending review" and no tier. Today 47 entries are in that state, 36 in bedding and 11 in clothing, and none of them is shown on a shelf. Thirty-eight supplement entries left that state on 2026-09-20 when they were reviewed and approved.

What changed this week. On 2026-09-15 we reviewed every flagged claim on the product catalog. Sixteen claims on thirteen products were dropped, attributed or cited. The check itself had three blind spots, which we fixed: it read "non-toxic" as a negation and never checked it, it accepted "no IEEE 1789 data" as evidence, and it accepted "certified organic" as naming a certifier. Twenty-nine health and efficacy statements in our own copy were rewritten, cited or removed, including a "clinically proven, 55% more plaque" line whose only source was a brand PDF. "Organic" was checked brand by brand across 119 entries: 16 now cite the certifier, and the rest say the word is the brand's.

How a listing gets on the map, and what "verified" means

The Local Food Map holds 16,641 listings: farmers markets, on-farm markets, groceries, CSAs, restaurants, meat producers and bakeries. They come from state agriculture registers, the USDA Local Food Portal, certifier directories, producers' own stockist pages, and operators' websites. A listing is a place a person can visit and buy food. A vendor with no location of its own is not listed.

Almost every listing carries a flag we call coord_verified. It means one thing: the map pin agrees with the listing's address as written. That check catches a bad geocode. It cannot catch a bad address. On 2026-09-14 we found three pins with that flag in the wrong place, because the source had filed an upstate New York address under the city "New York" and the pin landed, correctly, in Manhattan. We corrected them, and we now hold any row whose city is a state name or whose zip does not belong to its city. But the flag still means what it meant. It does not mean we confirmed the address with the business, that the business is at that address, or that it is open.

The site is checked before every publish for copy that claims more than this. Wording that calls the map or a listing verified fails that check, on the page and in the assistant's instructions.

What else a listing shows about its own provenance:

• Whether a website was matched to the listing, and at what level: to the street address (2,615 listings), to the city only (2,482), to a city name with no address to anchor it (84), or not found (11,460).
• Where the opening hours came from, and the date they were read, on the 3,738 listings whose hours we read from a source.
• For restaurants, meat producers and bakeries, an evidence status with five values, shown as written: confirmed with the owner or chef, published by the business, confirmed by staff, not independently confirmed, or sources disagree. Most listings on the map are "not independently confirmed", and the card says so.
• A line saying no website, social page or phone number was found, where that is the case.

A listing is removed as closed only on two independent signals, such as a dead website plus a press report or the operator's own successor page. A dead website alone is never enough. When a certifier directory turned out to carry a decade of dead entries, we delisted 17 confirmed closures on 2026-09-02, and we now run that check before any directory batch ships.

Rated, listed, and listed but not rated

Most map listings carry no rating at all. A rating on the map is separate from being on the map.

Restaurants are rated on cooking fats, sourcing, preparation, components made in house, and transparency, on a fixed ladder. Unknown cooking fats cap a restaurant at Tier 3. A kitchen cooking in seed oil and assembling rather than cooking is not listed. Thirty-two restaurants are rated today.

Groceries are rated at the chain level on three axes from the chain's own published policy: an ingredient standard (a published banned-ingredients list, a claim without a list, or none), a meat standard, and organic share. The source URL is recorded on each row. Both standards published earns Tier 1; one earns Tier 2; a claim with no list earns Tier 3; nothing on any axis means the chain is not rated. That is why every Whole Foods Market is Tier 1 and every Sprouts is Tier 3: the rating is of the chain's policy, not of the store on your corner.

Bakeries have a ladder, but no bakery has been asked the questions yet, so all 29 are unrated, and the card says so.

Listed, not rated is a separate state. A grocery we assessed against our standard and did not rate can still be worth finding, because a dated primary source names that location as stocking something that meets a ROOTED standard. Such a listing carries the badge "Listed, not rated", cannot carry a tier, and cannot carry any endorsing language. A check fails the publish if it does. Listing is not endorsement. Ten listings are in this state today.

The same state exists on the product side. Twenty-seven brand-level entries for meat, poultry and seafood sellers carry the "Listed, not rated" badge because their individual products have not been reviewed. They were shipped on 2026-09-03 with placeholder tiers and taken off them on 2026-09-15; a check now fails the publish if a brand-level row carries a tier.

Third-party claims

Some badges on the map report what a certifier or a producer says, not what ROOTED assessed. You can tell which is which because the badge names its source and says so: "Named in A Greener World's where-to-buy directory", "Certified by Real Organic Project, per its directory", followed by "Not a ROOTED rating", or "Separate from ROOTED's rating" on a listing that also has one. Every source that appears this way has to be registered in the site before its listings can ship. 1,380 listings carry such a badge.

On product entries, a claim we did not verify is written as the brand's: "Badger states", "brand-stated", "per the manufacturer's research".

This week we removed 1,048 sentences from listings because they named a brand, a certification or a percentage with no recorded source: a supplier claim repeated on 501 Whole Foods rows, a "63% of beef sales" figure repeated on 312 Sprouts rows, and 233 more across five chains. A sentence like that now fails the publish check when it appears on five or more listings without a source. A removed sentence comes back only with its source attached. One gap we know about: a claim reworded slightly on each row is not caught by that check.

What we get wrong, and how we fix it

We get things wrong. Addresses come from sources that make mistakes, businesses close, products reformulate, and our own checks have had bugs that we found and fixed in public.

If you find an error, write to team@rootedapproved.com. When a correction comes in, we treat it as a correction to our own data first, before we treat it as anything else. A corrected value is recorded alongside the value it replaced, so the change can be reversed if the correction turns out to be wrong. A listing is delisted only on two independent signals. Every change is committed with a dated note saying what changed and why.

We acknowledge a correction within 5 business days. Within 30 days we either publish the fix or tell you why we did not.

Rooted U corrections

This section lists what Rooted U got wrong: statements we published as fact that a source contradicts, or that said more than we could support in a way that changed what a reader would take away.

Each entry gives what we said, what was wrong with it, what the page says now, and the date the wrong wording left the page. Entries are permanent. If an entry itself needs correcting, a dated note is added beneath it. Nothing here is edited or removed.

In October 2026 we checked the sentences in Rooted U against primary sources, one sentence at a time. Of the 454 sentences in that check, 58 were supported as written, 237 were supported only in a narrower form, and 159 were removed (figures as of October 20, 2026). Seven sentences taken off the page on October 1, before that check began, are not counted in those figures. The entries below are the ones where what we had published was wrong.

Slide numbers count from 1 and give the slide's position when the wording was published. A staging id and a live sid name the sentence in our records.

How ratings stay independent

Where we have a commercial relationship with a brand, that relationship plays no part in how its products are rated or whether they appear here. Every product goes through the same standards, the same ladder and the same approval, whoever sent it and whatever else we do with them.

There are no brand submissions. A store guide lists only products we have rated, on the basis of a brand's or certifier's own where-to-buy page, and never ranks stores by how many rated products they carry.

What ROOTED is not

ROOTED is not medical, dental or nutritional advice. Talk to a clinician about your own situation.

Clothing on ROOTED covers men's, women's and unisex garments, rated on their materials and never on who wears them.

A tier is our opinion of a product against the criteria on this page, formed from what the brand publishes and what we could verify on the date we checked. It is not a statement of fact about a company's conduct, and it is not a test we ran on the product unless the entry says so. A map listing is a record of a place from the sources named on its card. It is not a promise that the place is open or that its address is right.

Last revised 2026-10-20.