A staged process for fair public-interest analysis, not a rush to publication.
Observed uses a staged process to move from an initial public-interest question to evidence checking, research development, benchmark comparison, right of response and publication review. The same process can identify concern signals, improvement signals or strong-practice signals where the public evidence supports it. Each stage is a quality gate, not just a task list.
Three questions guide the process.
The detailed work changes by matter, but the process always tests fit, evidence strength and publication fairness before any public-facing output is considered. That applies to accountability concerns, improvement patterns and positive-practice recognition.
Does the matter fit?
The first test is whether the issue fits Observed’s public-interest purpose, organisational focus, evidence model and ethical boundaries. This may include concern signals, improvement signals or strong-practice signals.
Does the evidence support it?
The second test is whether lawful public signals exist, whether they come from suitable source types and whether they support concern, improvement or positive-practice analysis.
Can it be published fairly?
The final test is whether the analysis remains source-cited, benchmarked, proportionate, response-aware and within the legal boundaries. Observed avoids unsupported criticism and unsupported praise.
Observed can identify concern, improvement or strong practice.
Accountability is not only about failure. A disciplined public-interest model should also be able to recognise responsible conduct, clear governance, benchmark alignment and meaningful improvement when the public evidence supports that conclusion.
Concern signals
Public evidence may indicate a possible gap between organisational claims, stakeholder experience, recognised standards and observable conduct.
Improvement signals
Public evidence may show that an organisation is responding, adapting, strengthening governance or improving practice over time.
Strong-practice signals
Public evidence may support recognition where organisational behaviour appears aligned with academic benchmarks, good-practice standards and public-interest expectations.
Not every signal becomes a report.
A careful process needs clear stop points. Observed may proceed only where the matter fits the purpose, the evidence supports the analysis and publication remains proportionate, whether the finding is critical, neutral or positive.
From public-interest question to publication decision.
The process applies to platform-led research, client-initiated requests and commissioned research. The same evidence, ethics, privacy, legal and human-review gates apply whether the output identifies concern, improvement or strong practice.
Initial question and fit assessment
This stage determines whether the issue belongs within Observed’s purpose and evidence model. It screens for public-interest relevance, organisational behaviour focus, plausible public evidence, bad-faith motive, privacy risk and safety risk. Positive-practice requests are screened with the same discipline as concern-based requests.
Evidence check
This stage tests whether publicly available signals exist and whether they are sufficient to justify deeper benchmark analysis. It looks at source types, signal categories, source diversity, evidence gaps and balancing material, including material that may support concern, improvement or strong-practice findings.
Research development and information-gap work
This stage strengthens the evidence base before full analysis. It may include deeper source research, timeline development, source registers and information-gap work. Public-information request drafts, including official information request pathways, belong here where relevant. The aim is to test the pattern, not to force a negative or positive conclusion.
Benchmark comparison and pattern analysis
This is the core analysis stage. Classified public signals are compared against peer-reviewed academic frameworks and recognised good-practice standards. The analysis identifies alignment, divergence, confidence ratings, limitations, accountability questions, improvement indicators and strong-practice indicators where supported.
Right of response and publication review
This stage ensures that a named organisation has a fair opportunity to respond before publication where required. Proposed findings or questions are shared with an appropriate organisational contact, and the response is considered before any publication decision is made. This applies to critical findings and to positive-practice recognition where the organisation should have an opportunity to confirm, clarify or correct the public record.
Publication and distribution
This stage publishes and distributes findings in formats appropriate to the evidence, audience and public-interest purpose. Publication may include a report, article, executive summary, stakeholder briefing, media backgrounder, infographic copy, recognition profile or distribution pack.
Publication requires more than a signal.
The process is deliberately built around safeguards that prevent weak, narrow or disproportionate material from becoming public-facing analysis. The same safeguards apply to concern-based reports and positive-practice recognition.
The issue must concern organisational behaviour, public accountability, stakeholder harm, public trust, improvement or strong public-interest practice, not a purely private dispute.
Public signals must be lawful, attributable, relevant and strong enough to justify analysis or recognition.
Privacy obligations are considered against the New Zealand Privacy Act principles.
No named-organisation output leaves the system without human review and sign-off.
The process can stop, narrow or change direction at any point.
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. Positive-practice recognition is also withheld unless public evidence, benchmark alignment and source consistency support it. 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.
Observed publishes comparisons, not verdicts or endorsements.
Published content must remain comparative, source-cited, benchmarked, confidence-rated, proportionate and legally reviewed where required. It may identify risk, improvement or strong practice, but only within the limits of the evidence.
It must not become unsupported allegation, reputational attack, revenge content, inflammatory campaign material, personal targeting, unsupported promotion or content that goes beyond the available evidence.
The organisation may be named. The verdict is not asserted. The benchmark does the work.
Have a public-interest question that may fit the process?
Observed can assess whether an issue, concern, improvement pattern or example of strong practice has public-interest relevance, a lawful public evidence base and enough source diversity to justify deeper research.