PeopleFirstVerified
Scoring methodology
How PeopleFirstVerified assesses the visibility, clarity and completeness of information found on public websites and in publicly linked documents.
PFV Public Evidence Methodology v1.0What the assessment measures
PeopleFirstVerified assesses the visibility, clarity and completeness of relevant information found on publicly accessible website pages and linked documents.
The resulting report is a dated snapshot of what an external visitor could reasonably identify at the time of review.
The assessment does not examine internal systems, test whether published commitments operate in practice, determine legal compliance, or constitute certification or legal advice.
Assessment areas and weights
Criterion ratings
Artificial intelligence classifies the extracted public evidence against fixed assessment criteria. Each applicable criterion receives one rating.
| Rating | Meaning |
|---|---|
| 0 | No relevant public information was identified. |
| 1 | Information is limited, vague, incomplete or difficult to find. |
| 2 | Information is partial but reasonably clear. |
| 3 | Information is clear and substantially complete. |
| N/A | A conditional criterion is not applicable because no positive public evidence indicates that the relevant AI activity occurs. |
Only conditional AI-related criteria may be excluded as not applicable. A business is not penalised merely because no public evidence of AI use is found.
Assessment criteria
Transparency
- T1: Organisation and service clarity
- T2: Public disclosure of relevant AI use, where applicable
- T3: Explanation of AI roles, decisions and limitations, where applicable
- T4: Relevant policies and public disclosures
- T5: Applicable Article 50 disclosures, where supported by positive evidence
Human access
- H1: Identifiable human contact route
- H2: Information about obtaining human assistance
- H3: Published complaints or escalation route
- H4: Human consideration of significant AI decisions, where applicable
Accountability and governance
- A1: Organisational identity and contact information
- A2: Relevant privacy and data-responsibility policies
- A3: Published AI ownership or governance information, where applicable
- A4: Route to question significant AI-influenced decisions, where applicable
- A5: Published AI oversight or review arrangements, where applicable
How scores are calculated
For each section, the points earned are divided by the maximum points available for its applicable criteria and converted to a percentage. Criteria rated N/A are excluded from both figures.
Scores are rounded to the nearest whole number by the PeopleFirstVerified application. Artificial intelligence does not calculate or select the section or overall scores.
Score bands
| Score | Interpretation |
|---|---|
| 80–100 | Strong public evidence |
| 60–79 | Adequate or promising public evidence |
| 0–59 | Public evidence needs improvement |
| No score | Insufficient evidence to assess any applicable criteria reliably |
Article 50
Article 50 considerations affect the assessment only where positive public evidence indicates a covered AI use and a potentially relevant provider or deployer role. The absence of a blanket statement about AI use is not itself treated as a gap.
The assessment does not determine whether Article 50 legally applies or whether an organisation complies with it. Where relevant public information is not found, the report describes a public-evidence gap that may require separate qualified legal or technical review.
Use of artificial intelligence
This is an AI-assisted public-evidence assessment. Website pages and linked documents are collected automatically. AI classifies the extracted evidence against the published criteria and generates the narrative findings.
The PeopleFirstVerified application calculates the section and overall scores separately using the fixed formula described on this page. AI does not determine the final scores.
The preliminary report is not independently reviewed by a human.
Important limitations
The report contains text generated using artificial intelligence.
Missing public evidence does not prove that an underlying policy, safeguard or practice does not exist. Scores do not measure the organisation’s internal conduct, AI safety, trustworthiness, overall responsibility or legal compliance.