Every SINK score is produced by the same formula, using the same rubric, applied to publicly available data. No exceptions. No special treatment. This page explains exactly how it works so you can verify any score yourself.
SINK scores are physics-first climate assessments built entirely from public data, on an open methodology (v2.5.10). Companies cannot pay to change a score — the £99 Enhanced Assessment deepens evidence, it never buys a number.
SINK is a physics-first environmental impact score. We measure what the planet physically experiences — absolute emissions, ecosystem damage, resource depletion, water impact — not reporting maturity or corporate process. Scores are independent assessments open to challenge and correction by anyone with evidence.
Transparent. Reproducible. Open to challenge.
SINK = (0.3 × Base + 0.7 × Performance) × ScaleIndustry sustainability ceiling. Oil & Gas: 20. SaaS: 75. Fixed per sector — it caps how high a company in that industry can score.
Ten questions, same rubric for everyone. Emissions, targets, governance, offsets, lobbying. Where real effort shows.
Total emissions volume. 50M+ tonnes triggers a 0.60× multiplier. Physics doesn't negotiate — size is part of the score.
Here's how that formula produces Apple's score of 42.
Electronics / Hardware sector ceiling, derived from lifecycle emissions and recyclability benchmarks.
10 rubric questions: emissions targets, Scope 3 disclosure, governance, offsets, lobbying.
Weighted average: base impact has a 30% floor, performance drives 70% of the result.
Apple's 20M+ tonne Scope 3 footprint triggers the multiplier. See methodology §4.
60.3 × 0.70 ≈ 42. Below expectations.
Every company is assessed against these ten questions. The first seven measure physical impact — what the planet actually experiences. The last three measure whether the company is improving and can be held accountable. Performance = sum of all ten scores.
Scores between anchors (0, 3, 6, 10) reflect how closely a company meets the next threshold. Operational evidence (energy bills, supplier records, water bills) is accepted at all levels — formal sustainability reporting is not required to score well.
Each industry gets a base score reflecting the physical environmental impact of its core business activity. Click any row to see the rationale. All scores are open to challenge.
New sectors added as companies are scored. Base scores set by editorial review and open to challenge with evidence.
The planet experiences absolute tonnes. Total emissions volume applies a multiplier between 0.60× and 1.00× to the weighted score.
When we cannot obtain a sourced emissions figure for a company, it is still scored — not set aside — with a conservative 0.70× scale penalty (the harsh end of the table). We distinguish two cases and never overstate the stronger one: not retrieved means a search did not surface a citable figure — this is not a claim that the company fails to disclose; no public figure is reserved for confirmed non-disclosure. The one exception is a genuine small company — fewer than 250 employees and revenue under £50m — where the figure is treated as an estimate rather than penalised as a disclosure gap. This exception never applies to heavy sectors (oil & gas, mining, coal, chemicals, cement/glass) or to financial firms (banks, insurers, investment trusts, asset managers), whose material emissions sit in what they finance or hold rather than office headcount; nor when revenue is unknown.
Theoretical near-maximum. Requires exceptional performance across all dimensions at scale.
Best-in-class. Only a handful of companies in our database have reached this level.
Genuine sustainability leadership backed by verified data and science-based targets.
Meaningful effort with room for improvement.
Some action but significant gaps remain.
Major disclosure or performance deficiencies.
Minimal effort or actively harmful practices.
We mark hard, and we mark everyone the same way. Scores above 75 are exceptionally rare — reflecting how demanding the rubric is. The majority of companies score between 30 and 60. A score of 60+ represents genuine sustainability leadership, the kind backed by verified data, science-based targets, and transparent governance.
A separate designation for companies whose core business demonstrably removes more greenhouse gas than it creates. The badge is displayed alongside the SINK score — it does not change the score itself.
Eligibility requires all three:
Companies cannot pay to receive this designation.
Every score publishes as Pending Review and stays open to challenge. Anyone can contest any question with evidence, for free. A score becomes Verified only after editorial review of sustained community evidence and a stable score over time — it is never automatic.
The default state for every score. Scored using publicly available data and published straight away, open to challenge by anyone with evidence.
Confirmed by editorial review after sustained community evidence and a stable score over time. A deliberate human step, not triggered by a vote count.
We score from public evidence, which is the right starting point because it is checkable and anyone can contest it. It has one real weakness: a company that does good work quietly scores below a company that does average work loudly. That is a reporting artefact, not a judgement on practice.
Scoring low on something you actually do well? How companies can show their evidence →
Every score is expected to cite at least eight independent sources. For companies with limited public sustainability disclosure — typically smaller or privately held companies — that threshold is sometimes unreachable. Where a company's public sustainability disclosure cannot meet the standard source threshold, the score is published with the complete verified source set disclosed rather than padded. An editorial note on the score record identifies this situation and lists the sources used.
An audited annual report filed at Companies House counts as one of those independent sources. Audit assurance is what independence is really a proxy for: a company's own website is discounted because nobody checked it, while a filed report has been audited and lodged with the statutory registrar. A filing contributes exactly one independent party however many filings are cited, and it is never enough on its own — the source count and the limit on how much can come from a single domain still apply.
Where a company files its annual report as a scanned image, we read the filing by optical character recognition so its statutory carbon disclosure can be quoted and cited like any other source. A company is not marked down because of a document we could not parse.
Some scores publish as provisional evidence. Those are scores where every quality check passed — every source live, none from a data aggregator, none resting disproportionately on a single domain — but there was simply less evidence available than our full-confidence threshold of eight sources from three independent parties. Those pages carry a banner stating the actual counts, and a note saying what is missing and what would close the gap. The threshold has not moved; what has changed is that falling short of it now means publishing with the count shown rather than not publishing at all. Thinly reported is not the same as badly reported, and a reader is better served by the number and its basis than by silence.
A company's sector sets its base score, selects which rubric applies, and defines its peer group — so a misclassification is three errors, not one. Sector is therefore resolved from external reference sources before a company is scored, in a fixed order of authority:
Where sources agree, the company is scored. Where a higher- or equal-authority source contradicts the primary one, the company is held for review rather than scored on a coin-flip; a lower-authority source that disagrees is recorded as noted dissent but does not block. A company with no external classification source at all is flagged accordingly.
Every classification carries the source it came from, the raw code or label that source assigned, and the date it was determined — so a company that disputes its sector can see exactly how it was reached.
Every question on every score has a “Challenge” button. Click it, select the tier you think is correct, provide your evidence — a source URL is required — and submit.
Challenges are published publicly. When community evidence supports a more accurate assessment, the score is updated, the challenger is credited, and the change is logged permanently in the score history.
We'd rather be corrected than wrong. That's the point.
New schema across all company pages: source-tracking per question, editorial summaries, per-question evidence paragraphs. Rubric structure unchanged; lobbying remains assessed within Controversies (Q10) as a deduction factor.
Published rationale for every base impact score. SaaS revised 75→70, Electronics 45→40, Apparel (Durable) 25→30. Added interpolation note and verification source requirements.
Formula, 10-question rubric, base impact table, scale penalties, scoring bands, verification tiers, and challenge process.
Every score on the leaderboard shows its full breakdown — base, performance, scale penalty, and the 10 question scores with evidence and sources.