Our World in Data

Our World in Data (OWID) gathers statistics on poverty, health, energy, the environment, education, democracy, and dozens of other global topics, then cleans and harmonises them into consistent, comparable long-run series. Each topic pairs plain-English writing with interactive charts, and every figure traces back to the official source it was drawn from, so provenance is never a mystery.

For builders, the appeal is the packaging. Every chart can be downloaded as a ZIP holding the data as a CSV file, the metadata in JSON format, and a README that explains what you are looking at. The same data is reachable through an API and a public catalogue, and there is an official Python package for pulling series directly into a pipeline, which beats scraping charts by hand.

Turning this into RAG takes a little thought, because it is tabular rather than prose. The strongest pattern is to pair each series with its metadata and the surrounding OWID article text, so a retrieved chunk carries not just the numbers but what they mean, how they are defined, and where they came from. Done well, you get a knowledge base that answers a question like how child mortality has changed since 1990 with a figure and a citation rather than a vague sentence.

This is the cleanest, best-documented statistics source we list, which makes it an excellent starting point whenever you need trustworthy numbers with clear sourcing: fact-checking assistants, research tools, and any system that must not invent figures. Keep two things in mind. OWID aggregates other people's data rather than collecting its own, so cite the original provider it names for anything sensitive, and the series carry the usual reporting lags, so they are current to within a year or two, not to today. Everything is released under CC BY 4.0, so commercial reuse is fine as long as you attribute. It complements World Bank Open Data, Eurostat, and UN Data, often presenting the same figures in a far more usable shape.