How controversy scores are calculated
The four-dimension methodology behind controversy scores — scale, severity, proximity and company response — the formula that combines them, and a worked example.
Every incident found by Real-time Controversies receives a single controversy score from 1.5 to 10 and a four-tier severity label. The score is built from four dimensions, each scored with written reasoning, so you can always see why an incident scored the way it did — not just the number.
The four dimensions
Scale — how many were harmed
The breadth of parties materially harmed — people, animals, ecosystems, investors, consumers or market participants — counting only harm traceable to this specific company.
| Score | Label | Meaning |
|---|---|---|
| 1 | Negligible / administrative | No identifiable harmed parties; impact purely procedural or reputational |
| 2 | Contained | Partial impact at a single, bounded location |
| 3 | Regional | Multiple sites or communities — broader than one location |
| 4 | Widespread | Cross-border breadth, or harm to the general public at large |
Severity — how deep the worst harm was
The depth of harm for any affected party, regardless of how many were affected. It is scored on the worst harm evidenced — a single irreversible instance is enough for the top score.
| Score | Label | Meaning |
|---|---|---|
| 1 | Minor / procedural | No material harm; purely regulatory, administrative or reputational |
| 2 | Moderate / recoverable | Real but temporary harm; affected parties return to baseline |
| 3 | Serious / lasting | Lasting harm that stops short of death or total destruction |
| 4 | Grave / irreversible | Harm that cannot be undone — a single confirmed instance justifies a 4 |
Proximity — how directly the company caused it
Scored for the named company only, not third parties or industry peers. Proximity acts as a multiplier on the score.
| Level | Multiplier | Meaning |
|---|---|---|
| Third-party supply chain | ×0.75 | Harm connected through a distant business relationship the company does not control |
| Direct supply chain | ×0.875 | The company's own decisions enable harm through a direct business partner — a supplier, contractor, franchisee or joint venture |
| Own operations | ×1.0 | The company causes harm through its own operations, employees or controlled entities |
Company response — how the company reacted
Only actions by the company itself are evaluated — not regulators, courts or third parties. When several behaviours are present, the least accountable one applies. Response also acts as a multiplier.
| Response | Multiplier | Meaning |
|---|---|---|
| Active remediation | ×0.75 | Goes substantially beyond the specific finding to address root causes — for example auditing other operations or suppliers beyond the one implicated |
| Acknowledgement | ×1.0 | Accepts accountability — apologises, cooperates with investigations, fixes the identified issue |
| Passive compliance | ×1.0 | Does the legal minimum only — pays the fine or withdraws the product, with nothing voluntary beyond it |
| Unknown | ×1.0 | No public response, or only non-committal language with no concrete committed action |
| Resistance and denial | ×1.125 | Openly disputes findings, appeals rulings or defends the practice — without deception |
| Cover-up | ×1.25 | Actively suppresses evidence, deceives, or silences witnesses in response to this specific controversy |
The formula
controversy score = (scale + severity) × proximity multiplier × response multiplier
Scale and severity each contribute 1–4, so the raw sum ranges from 2 to 8. With the multipliers applied, the effective range is 1.5 to 10. The absolute worst score of 10 requires grave, irreversible harm at widespread scale, caused by the company's own operations, together with an active cover-up — a combination that is deliberately rare.
Severity labels
| Label | Score range |
|---|---|
| Low | ≤ 3.0 |
| Moderate | 3.1 – 5.0 |
| Severe | 5.1 – 7.0 |
| Very severe | > 7.0 |
Every score also carries a confidence rating — low, medium or high — reflecting how well the available evidence supports scoring this specific company across all four dimensions.
Tuned to each of the 24 themes
A single definition of "grave harm" does not work: a data breach and a chemical spill reach the top of the scale for completely different reasons. Scale and severity are therefore calibrated separately for each of the 24 ESG themes — what counts as "widespread" for a greenwashing incident reflects how broadly the misleading position is embedded in the business, while for a health-and-safety incident it reflects the people harmed. Proximity and company response use the same definitions across all themes.
This is what keeps scores comparable across the whole dataset: a "severe" greenwashing score and a "severe" health-and-safety score each reflect what actually matters in that area.
Scores evolve with the story
An incident's score is a single running score, not a snapshot. When new coverage arrives — a fine is issued, a class action is filed, the company responds — the incident is re-assessed, and only the dimensions the new information materially changes are revised. Severity and status are tracked over time, so an incident's history shows how the story developed.
A worked example
The UK Advertising Standards Authority ruled that Volvo Cars misled consumers about the stated range of the EX30 electric vehicle.
| Dimension | Assessment |
|---|---|
| Theme | Greenwashing |
| Scale | 1 — no identifiable materially harmed parties |
| Severity | 1 — no material harm; a communications adjustment was required |
| Proximity | Own operations (×1.0) — the company's own advertising claims |
| Response | Acknowledgement (×1.0) — the ruling was accepted |
| Confidence | High |
(1 + 1) × 1.0 × 1.0 = 2.0 → Low
A real regulatory ruling that still scores Low — because the score reflects the actual harm evidenced, not the drama of the headline.
Common questions
Why did a big news story score Low? Media attention is not an input to the score. An incident scores on the harm evidenced, how directly the company caused it, and how the company responded. A widely covered story with no material harm scores low; a barely covered story with irreversible harm scores high.
Can the severity bands be adjusted? The underlying dimension scores and formula are fixed, but the banding thresholds can be adjusted per client to match your risk framework.
Where does the reasoning come from? Every dimension score is delivered with a written explanation and the supporting source articles, so you can see the evidence behind each rating — not just the final number.
Does a company's good behaviour elsewhere offset a controversy? No. Each incident is scored on its own evidence. A company's response to that specific controversy is the only conduct that moderates the score.