How Observed turns public information into balanced accountable insight.
Observed works from lawful, public and attributable information. Sources are classified, weighed against their limits, compared with recognised benchmarks and reviewed before any publication decision is made. The same method can identify concern signals, improvement signals and strong-practice signals where the evidence supports them.
Public evidence. Benchmark comparison. Pattern analysis.
The methodology keeps the work comparative and evidence-led. It does not convert a concern, positive claim, review or report into a finding without source classification, benchmark fit and human review.
Public signal collection
Observed identifies publicly available material linked to organisational behaviour, governance, culture, stakeholder experience, public claims or accountability.
Each signal is recorded with its source type, access context, relevance and limitation.
Benchmark comparison
Classified signals are compared against peer-reviewed research, recognised professional guidance and good-practice frameworks.
The standard exists independently of Observed. The work applies it, explains it and states its limits.
Pattern analysis
Outputs identify where public signals appear to align with risk indicators, diverge from good practice, show improvement over time, or demonstrate strong alignment with accountability expectations.
Every finding is source-cited, benchmarked, confidence-rated and limited.
The method tests what the evidence supports, not what the request expects.
Observed may identify concern signals where public evidence points to risk, inconsistency or accountability questions. It may identify improvement signals where public material shows a credible change in practice over time. It may also identify strong-practice signals where organisational conduct, public communication, governance or stakeholder outcomes appear to align with recognised benchmarks.
Positive findings are not awards, endorsements or promotional claims. They are treated as benchmarked observations, with the same source diversity, limitation statements, confidence ratings and human review used for critical findings.
A signal is not a finding. It is a classified input.
A public signal is lawful, accessible and attributable information that may indicate something relevant about organisational conduct, governance, culture, public claims, improvement, strong practice or stakeholder experience.
Signals are not treated as automatic proof. They are classified, compared with other source types, checked for contradiction and assessed against benchmarks before any finding is developed.
Different source types carry different weight. Formal records, reported information, anonymous employee-experience signals, organisation-owned material and stakeholder material are not treated the same way.
Regulator notices, tribunal decisions, public registers, funding records and public accountability material.
Classified as reported public information, not automatically treated as verified fact.
Anonymous employee-experience signals, not enough by themselves for named-organisation findings.
Websites, annual reports, impact reports and social media treated as public claims or positioning.
Public submissions, reports, parliamentary material and official information responses assessed by context.
Research frameworks and recognised guidance used to support comparison and interpretation.
The standard comes from research, not from Observed.
Observed compares public signals against a curated benchmark library of academic frameworks, professional standards and public-sector guidance. Benchmark choice depends on the issue, sector, evidence base and publication risk.
Workplace harm
Workplace bullying culture, psychosocial risk and psychological safety benchmarks are used where public signals relate to workplace harm, culture risk or employee experience.
Relevant references may include WorkSafe New Zealand psychosocial risk guidance and Employment New Zealand bullying guidance.
Governance and accountability
Governance benchmarks are used where public signals relate to board oversight, public claims, funding accountability, leadership responsibility or organisational control.
Relevant references may include Institute of Directors New Zealand governance guidance.
Public-interest reporting
Public-interest benchmarks are used where the matter involves public records, official information pathways, public funding, stakeholder trust or accountability settings.
Relevant references may include Ombudsman guidance on official information requests.
Evidence standards prevent overclaiming in both directions.
Standards are not just content rules. They decide whether a matter can proceed, whether it must be narrowed and whether publication is fair. The same discipline prevents unsupported criticism and unsupported praise.
Observed does not use surveillance, impersonation, hacking, private databases, covert recordings or unlawfully obtained material.
Sources must be identifiable enough to record where they came from, when they were accessed and how they should be classified.
Each source is first treated according to its type, role, context and limitation before being used in analysis.
AI may assist with organisation and comparison, but human review governs interpretation, publication and language.
Named findings require source diversity.
Named-organisation findings are suppressed unless signals are drawn from enough independent source types. Multiple anonymous reviews alone do not meet the threshold, regardless of volume. Where the threshold is not met, the matter may be held, declined, referred elsewhere, used only as background context for a sector-level pattern, or added to a watchlist.
The analysis must stay proportionate.
Privacy obligations are considered against the New Zealand Privacy Act principles. Named individuals are removed from outputs where appropriate, and small-sample identification risk is considered before publication.
Published content must remain comparative, source-cited, benchmarked, confidence-rated, proportionate and legally reviewed where required. It must not become unsupported allegation, reputational attack, revenge content, personal targeting or unsupported promotional praise.
The organisation may be named. The allegation is not made. The benchmark does the work.
Have a public-interest question that may meet the method and standards?
Observed can assess whether an issue, improvement story or strong-practice example has public-interest relevance, a lawful public evidence base and enough source diversity to justify deeper research.