We distinguish what happened from what changed.

Four different kinds of statement

Activity

An activity records what the Programme or another responsible actor did: for example, a session was offered, a document was reviewed or a handoff was initiated. Activity does not show that the intended person could use it or that anything changed.

Output

An output records an immediate product or completion event: for example, an accessible version was produced, a decision was recorded or an original issuer issued a defined record. An output should identify the owner and acceptance criteria. It is not automatically an outcome.

Outcome

An outcome describes a defined change for a person, institution or process over a stated period. It requires a comparison, source and method capable of supporting that interpretation. A reported experience, an observed behaviour and an institutional decision are different forms of evidence and should not be collapsed.

Longer-term effect

A longer-term effect concerns change that extends beyond the immediate pathway and may have many causes. Claims at this level require particular care with attribution, alternative explanations, time, missing data and unintended consequences.

OECD evaluation guidance emphasises that evaluation criteria should be used thoughtfully and in context, rather than mechanically.¹⁰ The OHCHR indicators framework similarly supports clear definitions, data sources, methods and limitations.¹¹ These references inform the reporting discipline; they do not demonstrate any AZGBP result.

Every indicator needs a dictionary entry

Before a figure appears, the controlled record should identify:

• the question the indicator is intended to inform;
• the exact numerator, denominator and unit;
• inclusion and exclusion rules;
• source and collection method;
• period and cut-off date;
• responsible owner and reviewer;
• missing, unknown and suppressed values;
• disaggregation and protection rules;
• baseline or comparison, where relevant;
• known bias, uncertainty and limitation;
• correction and version history;
• the decision the evidence may inform.

Changing a definition creates a new version. A trend should not bridge incompatible definitions without an explicit explanation.

Evidence should support a decision

The learning cycle asks what should happen next:

Continue

Continue only where the pathway remains authorised, the controls are functioning and the available evidence supports continuation within the defined scope.

Change

Change the wording, process, accessibility, information request, role allocation or technical implementation when evidence identifies a correctable weakness. Record what changed and how the effect of that change will be reviewed.

Pause

Pause new action when evidence, authority, safety, quality, data or operational reliability is uncertain enough that continuing would be irresponsible. A pause is a control state, not a concealed failure.

Stop

Stop when the pathway no longer has a valid purpose or authority, creates unacceptable risk, cannot meet its conditions or is not supported by the evidence required for continuation. Preserve appropriate records and communicate the status honestly.

Listen without turning testimony into a metric

People may describe experiences that a numerical indicator misses. Qualitative evidence should preserve context, consent, uncertainty and diversity of experience. It should not be reduced to a decorative quotation or treated as representative of everyone.

Where accounts are published, the rights, safety, identity treatment, translation, editorial changes and withdrawal limits should be documented. A compelling story cannot substitute for systematic evidence, and a dataset cannot erase an individual experience.

Publication standard for a report

An approved report should state the question, method, period, population, definitions, sources, missingness, exclusions, privacy treatment, findings, uncertainties, limitations, review status, corrections and responsible publisher. It should distinguish observation from interpretation and correlation from causation.

The report page should expose material limitations before or beside the key finding. A downloadable file should match the visible version and carry the same status and correction record.

Current evidence state

No programme-specific findings or results are approved for publication in this draft. The absence of figures is intentional. Demonstration counters, invented testimonials, synthetic success rates and unsupported scale statements must not be used to make the page appear complete.

Empty and disabled states

No approved reports

Programme services are introduced in defined phases. Before any public action becomes available, its responsible entity, authority, information safeguards, security controls and support route are confirmed and presented alongside it.

Data unavailable

Programme services are introduced in defined phases. Before any public action becomes available, its responsible entity, authority, information safeguards, security controls and support route are confirmed and presented alongside it.

Report corrected

Programme services are introduced in defined phases. Before any public action becomes available, its responsible entity, authority, information safeguards, security controls and support route are confirmed and presented alongside it.

Closing actions

View approved reports Review policies and controls