How the information the Council publishes is produced.

The portal's series are built on three kinds of source: the administrative register of enrolment and schools, the teaching staff register, and the population projections that serve as the denominator of the coverage rates. Every indicator states which of the three it comes from.

In this demonstration all three are simulated. In a real deployment each belongs to a different institutional system, with its own owner and its own closing calendar.

The mapping is the exception: the boundaries of the eighteen departments are real and come from geoBoundaries (gbOpen, ADM1), published under a CC BY 4.0 licence. They are used as they are, projected and simplified for screen; everything painted on top of them is illustrative.

Before publication, every series passes three checks: internal consistency (the disaggregations sum to the total), consistency over time (no year-on-year change is left without a documented explanation) and cross-checking between registers (enrolment against the school directory, and schools against teaching staff).

When a check fails, the series is not published: what is published is the note explaining why it is not there.

Collection is annual and follows the calendar the Council approves: a single cut-off date for every reporting organisation, a cross-validation window and a single publication date. The half-yearly bulletins open no new series: they comment on the existing ones.

Every national figure is published disaggregated by education level, sex, area, administration and department, and — where the source allows — by session and route. A national average that cannot be broken down territorially tells you nothing about a country as uneven as this one.

The coverage rates depend on the population projection used as the denominator: changing the projection moves the indicator without enrolment having moved. The retention rates depend on the register of transfers between schools. The infrastructure indicators are built on what the school reports, with sample verification afterwards.

These limitations are published next to each indicator, not in a footnote.

Every dataset is published with its dictionary: variable name, type, unit, permitted values and how missing data is handled. The dictionary is part of the download, not of a separate page.

The statistics module exposes a single query function. Replacing it with a call to the institutional enrolment service leaves the dashboards, the map, the filters, the tables and the downloads untouched.

The coverage rates need a demographic denominator. The integration for that is a projection service by single year of age and territory.

The student–teacher ratio and the accreditation figures are fed by the staff system, not by the school register.

News, documents, regulations, datasets, indicators, events and consultations are modelled as collections, ready to be administered from a CMS without touching any code.