Follow the evidence.
Hold business accountable.

Recognizability determines inclusion, not the rating. This starter list includes US and international companies but is not a worldwide popularity ranking. Headquarters does not tell you where a brand is sold.

Numeric scores are reproduced from WBA’s published 2026 social assessment. The breakdown and company scorecard explain those scores; headlines never automatically change them. We have not assigned confidence ratings for the independent conduct assessment, so confidence remains Not assessed. No scored evidence does not mean no harm.

Parent scores describe the assessed group, not each product, subsidiary or franchise. Ownership links and review dates appear in details. Independent conduct scores require a separate, documented assessment before they can appear here.

Historical backfill begins with a small, explicitly reviewed set. Automated regulator matches enter a separate lead queue; they do not become established cases. A reviewer must verify company identity, group related sources, distinguish allegations from outcomes, and record scope, rationale, response gaps, and remediation. Source and editorial revisions remain in the permanent history.

Case severity rubric v1: Limited means a bounded issue with limited documented impact; Material means meaningful consumer, worker, public-fund or community exposure; Severe means major or widespread harm or serious exposure involving vulnerable people; Critical is reserved for documented extreme harm such as deaths or systematic severe rights abuses. Not assessed is used when evidence is insufficient. Confidence describes support for the stated proceeding or outcome, not a probability that all allegations are true.

A scheduled page check does not renew an editorial review. Changed content creates a review task and may be a navigation change. Follow-up dates remain due until an actual review updates the case. Improvements require evidence of implementation, not merely a promise. A company-wide numeric conduct rating requires broader coverage and remains unassigned.

US enforcement and UK competition feeds are checked daily; the EPA facility-name search, UK OECD National Contact Point feed and existing case-source pages are checked weekly. The hourly scheduler runs only the checks that are due. This is limited US/UK coverage, not worldwide monitoring. EPA monitoring currently covers a Volkswagen facility-name query; ownership of matching facilities has not been established. Failed requests retry after at most one day and keep the previous evidence.

Headline matches are discovery leads. A publisher mention is not independent confirmation, and brand or subsidiary matches require responsibility checks. The display is limited to 100 leads; the history export retains every archived record and revision.

WBA excludes these companies from scoring because their available disclosure is in languages other than English. They are excluded from both rankings.

At launch, no company dimensions had been assessed. Current coverage is shown above. Existing watch allegations and automated headlines are review leads, not evidence of an established culture failure. Each dimension starts as Not assessed.

A reviewer may record Concern supported, Mixed evidence, or Practice supported only with a documented finding or verified outcome, a written rationale, the period and operations covered, dated sources, and a company response or an explicit note that it is missing. A policy statement alone is insufficient. These judgments apply to the cited scope, not every part of a company.

There is no composite score or weighting. Missing evidence stays unknown. Review dates remain visible; assessments older than 180 days are marked overdue. Hourly news collection does not constitute a new human review.

Background: Interview with Robert Jackall, European Management Journal (2006). Our rubric is independently designed and is not endorsed by the author.