How we rate brands

We research each brand from public sources: its own website and product pages, certification databases, company records, government and regulatory filings, credible investigative journalism and independent assessments. Each claim is checked against its sources and sorted into facts confirmed by primary or independent sources, claims the brand makes that could not be verified, contradictions, and gaps. The full research on each brand page lists its sources.

The scale runs from brands in it for the craft to brands built for speed and scale. A step describes what a brand puts first when quality and cost pull against each other. It is not a score of how good or bad a brand is.

First the evidence is scored on six things: materials, making, how long the products last, commercial pressure, scale and speed, and the owner’s influence. Fixed rules turn those scores into the range of steps the evidence allows: Artisan needs strong materials and strong making, for example, and Volume needs a business built on scale and speed. A judgement of the whole picture then picks the step within that range.

• What a brand does counts more than what it says. Claims alone can’t earn a step, and claims that couldn’t be verified never lower one.
• Certifications, audits and supplier lists are paperwork: having them never raises a step on its own, and not having them never lowers one.
• A tradeoff with a fair reason, like stretch fibres in activewear, isn’t held against a brand. Neither are savings from how the business runs, like selling direct or a lean team.
• Missing information never pushes a brand toward Commercial or Volume, and it can’t earn Artisan. Evidence that products last only ever helps.

The card groups what the research found into rows: Materials, Making, Built to last and The business. Under each row are a few lines on what the price pays for and where the brand saves to keep it down. The first three rows carry a verdict, the same score the step was placed on.

When the sources say nothing about materials or making, that row shows no verdict. Built to last appears only when there is evidence for it, and having none never counts against a brand. The business has no verdict: it shows how the brand sells and grows, and what else the price pays for.

A line crossed is documented serious harm in how a brand’s products are made: forced or child labour, deadly or unsafe conditions in its factories, workshops or suppliers, proven fraud or misrepresentation of certifications, or serious environmental violations in its production.

• It has to be established by an authority, a court, or credible investigative reporting.
• It has to be this brand’s own: it names the brand, its own factories or suppliers, or a supplier documented as making for it, or a ruling against its group explicitly covers the brand’s production.
• Older findings still count, with their year.
• Disputed, unresolved or still-investigated cases go in the portrait’s note, not under lines crossed.
• Harm away from making the products, such as an accident at an event or an executive’s personal wrongdoing, is not a line crossed.

A brand with any line crossed can’t be placed above Commercial. No lines crossed means none were found in public sources, not that there are none.

Who owns a brand is evidence of what drives it, not a verdict. Founder or family control, a stock-market listing or nonprofit ownership don’t limit the step. When a group or investment funds control a brand, the question is whether the owner changed the product or kept its making. A brand its group created as one of its labels, owned by a group whose main business is fast fashion or mass retail, or whose product changed under its owner (cheaper materials, faster releases, logo-led or licensed products) can be placed no higher than Commercial.

The owning group’s own record, such as a finding at a sibling brand’s suppliers, is not counted as this brand’s line crossed. The card can show it as context, attributed to the owner by name, and it never lowers the brand’s step.

AI models do the research, check the claims, score the evidence and write the portrait; code then applies the fixed rules above. We test the method against our editor’s own placements of a set of brands, but a brand’s portrait may not have been checked by a person, and it can be wrong.

The full research on each brand page shows the date it was last updated. It reflects what public sources said by then, so newer events may be missing.

The products on a brand’s page come from the brand’s own public online store, and each links back to it. We refresh them regularly. A brand can ask us to remove its catalog; its research stays up.

If you think we have something wrong, tell us. If you represent a brand, you can ask us to correct the research, add a statement, update the brand’s details or remove its catalog on the For brands page, or email support@stories-of-one.com.