Editorial standards
OfficialCost publishes financial, immigration, and education cost data. Because people make real decisions on these numbers, we hold the data to a documented standard. This page explains how a number gets published, kept current, and corrected.
We collect from official and authoritative sources — government statistics offices, central banks, international bodies (e.g. the World Bank and OECD), recognized open datasets, and vetted community submissions. Every source is registered, reviewed for legal use, and assigned a trust tier before any of its data can be published. See source methodology.
Values are extracted deterministically from stored evidence — never generated, guessed, or paraphrased by a language model. If there is no source for a value, there is no published value. The full pipeline is documented in how we verify data.
Each value is scored by a transparent confidence engine that weighs source trust, agreement between independent sources, and recency. A page is only published — and only indexed — once it clears coverage, confidence, and knowledge-density thresholds. Details: how confidence works.
When independent sources disagree beyond a variance threshold, we do not silently average them. The discrepancy opens a conflict that is held back from publication until it is resolved, so a contested number is never presented as settled.
Facts are immutable and versioned: when a value changes, we write a new version and supersede the old one, so the history stays auditable. A freshness score decays over time, and stale values are re-collected or flagged rather than shown as if they were current.
We aim to fix confirmed errors quickly. Because every fact displays its source and date, anyone can check a number and report a problem. To request a correction or add a data point, use the contribution form or email corrections@officialcost.com. Submissions are reviewed against the sourcing standards above before anything is published; accepted corrections create a new fact version with its own evidence.
Advertising and the paid API fund the project but never affect the data — no pay-for-placement and no sponsored rankings. More on this in about OfficialCost.