Institutional accreditation tests whether a defined language-education system works in practice. A policy can explain intent; it cannot, by itself, prove implementation or effectiveness.
Fix the inspection universe
Before sampling, identify the accountable legal entity, trading names, sites, languages, learner groups, programme types, delivery modes, staff roles and exclusions. The sample must represent material variation across that universe.
Triangulate each conclusion
A robust finding connects several forms of evidence: the stated rule, a traceable record, observed practice, stakeholder understanding and an outcome or monitoring signal. Contradiction between sources is evidence to investigate, not average away.
Lesson observation is sampled across teachers, levels, modes and learner groups. It evaluates planning, language use, interaction, adaptation, feedback and evidence of learning against shared criteria—not personal teaching style.
Classify and close findings
A nonconformity identifies the unmet requirement, objective evidence and affected scope. Closure requires proportionate correction, analysis of cause and evidence that the action was implemented effectively; a promised future action is not closure.