XBRL

XBRL (eXtensible Business Reporting Language) is a standardized format for tagging financial statement data so regulators, analysts, and software can read it consistently across filers.

How XBRL tagging works

An XBRL filing pairs each reported value with a tag drawn from a taxonomy, a published dictionary of reporting concepts. The tag identifies what the number means (revenue, total assets, deferred tax liability), alongside context information: which entity, which period, which currency, and which unit. That context is what lets software compare a figure across filers and across years without a human reading the statement first.

Taxonomies are jurisdiction and framework specific. A filer reporting under IFRS uses the IFRS taxonomy; a US filer uses the SEC’s. Where a company reports something the taxonomy does not cover, it can create an extension element, which is where comparability starts to erode. A filing dense with custom extensions is technically valid and practically harder to analyze.

Inline XBRL, usually written iXBRL, embeds the tags inside a human-readable HTML document, so one file serves both the reader and the machine rather than requiring a separate data exhibit.

Why XBRL breaks in practice

The common failure is not malformed XML. It is tagging that does not match the statement. A figure gets mapped to a slightly wrong concept, a sign convention is inverted, or a period context points to the wrong quarter. The filing validates, the rendered statement looks right, and the machine-readable version says something different from the page a person reads.

The second failure is process. Tagging is often done at the end, by a different team or an outside provider, working from a near-final draft. Late changes to the statement then have to be reflected in tags by hand under deadline pressure. Nothing in the format prevents this; it is a workflow problem that the format exposes.

How Rexfin handles it

Rexfin’s extraction layer reads structured and semi-structured filed documents, including tagged formats like XBRL, as one more source type feeding the same reconciliation and citation chain. A tagged concept becomes a mapped, reconciled input rather than a number retyped into a model, and the figure that appears in a report keeps its link back to the filing it came from. That is the same lineage treatment applied to statutory reporting generally.