Methodology
How we rate
What a rating means, what goes into it, where the information comes from, and how we keep money out of it.
What the labels mean
Every brand we've assessed gets one of the labels below. There's no number, just the rung.
| Label | What it tells you |
|---|---|
| ExcellentEarned | Strong across the board, usually backed by certifications or an independent audit. Rare. |
| GreatEarned | Specific, detailed practices across environment, labor and how the company is run, usually backed by certifications. |
| Good | Real, specific commitments with evidence behind them. |
| Fair | Some good practices, some gaps. |
| Poor | Little specific evidence of good practice, plus at least one reported concern we count, or a brand contradicting its own commitments. |
| Limited info | We looked, but found too little public information to rate the brand. Not a judgement of its practices. |
| Rating under review | Our checks found at least two recent, serious controversies. We don't show a lowest rating while we re-check that evidence. Thin information, or a brand contradicting itself, never puts a brand here on its own. |
| Rating pending | We haven't assessed the brand yet. Not a bad mark. |
Every rating is labelled "Likely" (for example, Likely Good) because it's our best read of the public record, not an audit. Brand pages also show how confident that read is.
A rating is our opinion, based on the public sources we cite. Leif List is not a certification, testing or verification body. We don't audit factories, test products or confirm facts with the companies we rate. What that means in practice.
What goes into a rating
Three pillars, each scored from its own evidence, then blended into one rating.
| Pillar | What it covers |
|---|---|
| Environmental | Materials, emissions and energy, packaging and waste, sourcing. |
| Labor Rights | Supply-chain transparency, forced and child labor, worker treatment. |
| Compliance | How responsibly the company is run and how openly it reports. |
Within each pillar, three kinds of evidence count:
- How specific the brand is. "We care about the planet" earns nothing. Named materials, published supplier lists and measurable targets do.
- Specific commitments. Only ones about real environmental or labor performance. Paperwork every public company files anyway, like a whistleblower policy or a board-diversity statement, doesn't count.
- Certifications. A certification confirmed in the certifier's own registry counts more than one the brand only claims. Some, including ENERGY STAR, Leaping Bunny, USDA Organic, EPA Safer Choice and 1% for the Planet, are shown on the brand's page but don't change the rating. Signatory lists, management-system certificates (like ISO 14001) and index listings don't count at all, because they don't measure outcomes.
On a brand's page you can open each pillar to see the evidence underneath it. If you set your own values, results are re-weighted toward what you care about.
Why the top rungs are hard to reach
A big company with a disclosure team can publish a lot. That alone shouldn't let it outrank a brand that's actually doing better.
Excellent usually rests on certifications held across the business, or a high score on an independent audit like B Corp's. Great takes near-complete, specific evidence across all three pillars, and in practice that usually means certifications too.
A current, serious controversy can also keep a brand off the top rung, however good its paperwork is.
How we rate small businesses
A two-person workshop can't publish a corporate sustainability report. It shouldn't be marked down for that, so we look at what it actually does.
Smaller checklist. Most commitments we look for are written for large companies. For small and micro businesses we ask for fewer of them before awarding full marks, so a small brand isn't treated as secretive just because it's small.
Maker practices. Partner and Shop Local brands get a deeper, one-at-a-time assessment, and for small makers among them we also check five practices that show real impact at workshop scale:
- Recycled or upcycled materials
- Repairs or a lifetime warranty
- A take-back program
- Making products in-house, domestically
- Made to order, rather than overproduced
These count toward Good alongside regular commitments. A small maker that does all five can reach Great, and one that is also confirmed by two independent sources (a verified certification, or press, a supplier or an award on a site the brand doesn't own) can reach Excellent.
The bar is strict. Each practice needs a quote we've found on the brand's live website, not a summary. The brand also needs a clean record: no serious controversies and no claims that contradict each other. That way the rating is backed by something anyone can open and check, and it doesn't change between runs. Nothing a brand tells us privately counts.
Catalog brands are assessed in bulk and don't get this check yet. Bringing it to every small brand is on our list.
What pulls a rating down
- Serious controversies. Only high-severity findings affect a rating. Recent ones count most; older ones fade but never disappear unless they're resolved. A finding about a parent company counts for less than one about the brand itself.
- Contradicting itself. A brand whose public claims conflict with each other loses ground. That's greenwashing, and it's a fact about the brand.
Controversies are drawn from public reporting and shown on the brand's page as allegations, with their source. Findings from weak sources, like lawyer-advertising sites, review sites, AI-written wikis and social media, are hidden and don't count, and a finding without a source isn't shown as one. Other raters' scores, "no evidence of" notes and Prop 65 notices are labelled as such and don't count either. Not finding a controversy isn't the same as a clean record. Often nobody has reported on the company yet.
Where this falls short
Read every rating as an informed opinion with its sources attached, not a finding of fact.
- We're not a verification body. We don't audit factories, inspect supply chains or test products, and we don't confirm what we find with the companies we rate. A rating tells you what the public record shows, nothing more.
- The research is AI-assisted, and it can be wrong. The research agent can miss a source, misread a page or credit something to the wrong company. Quotes and links let you check every claim yourself, and our rules catch many mistakes, but not all of them.
- Coverage is uneven. Brands with a public record are easier to assess, so some categories are thin and some brands are missing. A missing brand is a gap on our side, not a judgment.
- Small brands publish less. "Limited info" means exactly that. It isn't a negative finding.
- Not every regulator is checked for every brand. We look for serious, well-sourced problems, but we don't search every agency's records for every company. Not finding a problem isn't proof there isn't one.
- Ratings describe companies, not products. A well-run company can still make a wasteful product, and a rating says nothing about whether a product is safe for you.
- Certifications are evidence, not character references. They may cover one product line or facility, and they lapse. We check them against the certifier's own registry where one exists, and some can't be checked.
- Public evidence lags reality. A company that changed course this year may not show it in the record yet.
- We get things wrong, and we want to know. If you think a rating is wrong, including about your own company, ask for a review. Send the source that shows what we missed. Requests come to our team and are reviewed, but we can't promise a reply or a timeline, and not every request leads to a change.
Where the information comes from
Product details come from public listings. A brand's record comes from a research agent that is asked to cite a source for every claim.
Products. The title, photos, rating and category come from the product's public Amazon listing or the partner brand's own store. We never touch your account, cart or orders. We only show a price when we can confirm it's current, so treat the retailer's checkout price as the final word.
The right company. A listing might name a marketplace seller or a sub-brand. We set aside sellers that aren't the maker, identify the real brand, and when a brand is owned by a larger parent company, may take the parent's record into account. If we can't confidently tell who makes something, the rating stays pending.
The research. An AI research agent asks every brand the same fixed set of questions. It's told to rely on three kinds of sources: the brand's own pages, certifier directories, and credible reporting such as news, regulators and court filings. Anonymous posts don't count. We ask it for a direct quote and a link for every claim, and we're adding checks that hold it to that.
The agent researches but doesn't decide. Ratings are calculated from its answers by fixed rules, so they can be re-derived and checked. Some ratings are set or confirmed by a person on our team after reviewing the evidence; when that happens, later automatic re-checks don't overwrite it.
What we double-check
Certifications are checked against the certifier's registry where we can (B Corp, USDA Organic, ENERGY STAR, EPA Safer Choice, Leaping Bunny, 1% for the Planet). If the registry shows a certification has lapsed or been revoked, we stop showing it on the brand's page. Claims that no registry covers go into a queue and are looked up by hand.
Serious controversies also go into a review queue, where a person reads the source. The finding is then confirmed, downgraded, marked resolved, re-attributed to a parent company or supplier, or removed if the source doesn't support it.
This is ongoing work, not a finished sweep. Both queues are worked in order of how much the answer would change a rating.
Product tags ("BPA-free," "organic," "plastic-free") are different. They're read off the listing and repeat what the seller says, with no one-by-one review. They power the filters, but they never move a brand's rating.
How we make money, and why it never touches the ratings
A brand can't pay for a better rating.
Leif List is free. We may earn a small commission, at no cost to you, when you buy through an Amazon link or from one of our partner brands' own stores. That's the only way we make money.
None of it influences a rating. Partner brands are rated on public evidence too, and being a partner can't buy a rung.
One thing does involve partners: on category pages and in search, the default order can move partner products up, up to about 40% of the spots, so you're not only seeing Amazon links. A partner only moves up when it's the same kind of product as the one it swaps with, only partners rated Good or better qualify, each brand is limited to a few spots, nothing is ever removed, and the rating on the card is always the real one. Choose Rating: High to Low to see products ordered by rating alone.