How our data works

Every series comes from an official statistical source: national statistical offices, central banks, and multilateral organizations such as Eurostat, the World Bank, the IMF, the OECD, and FAOSTAT. We keep each series source-native — provider names, series identifiers, dataset names, and citation titles stay exactly as the source publishes them — so you can always trace a number back to its origin. We do not scrape display labels or repackage third-party aggregations; rows start from provider-native identifiers.

A catalog entry and a chartable series are different things, and we label the difference instead of hiding it. Every series carries a readiness state: chart-ready means we hold fresh observations and you can plot it right now; still loading means the source lists it and our pipelines are hydrating its history; reference only means we index it for completeness but do not currently serve its observations. Counts shown on the site say which of these they measure. When a live count is temporarily unavailable, the page says “updating” rather than showing a stale claim.

Sources overlap: United States inflation exists in several official datasets at once. Library pages resolve that. Each page represents one economic concept in one country, with a stable address, and shows the best available series for it — switching sources underneath if a better one becomes available, without changing the page you bookmarked or cited. The alternatives remain one click away in the full catalog, with the exact source identity of what you are viewing always disclosed on the page.

Platform counts are recomputed from live data and stamped with when they were last verified — that is what the “Verified” badge on the indicators page means. Individual sources are checked against their own expected update cadence on the Data health board, which distinguishes operational, delayed, and degraded feeds. When a source stalls, the board says so; we would rather show you a delay than pretend one is not happening.

Each source’s license terms drive what we can show and how. Where a license requires attribution, the attribution is displayed with the data. Where terms restrict use, we restrict the surface — some sources appear in the catalog for reference without served observations, and sources whose terms we cannot satisfy are not loaded at all. Details are on the source policy page.