Public data is free. Usable data is not.
The numbers are public. The work of making them chartable is the actual product, and it is why a public-data business can exist at all.
Everything on Kitegraph's statistic pages comes from somewhere you could go yourself. The World Bank publishes its indicators. FRED serves half a million series. The Treasury, the EIA, the Census Bureau: all public, all free, all one HTTP request away. A reasonable person asks what a company adds on top of that.
The answer is that "public" and "usable" are separated by a large amount of work that never shows up in the chart.
Every source is its own small country
Public data comes from a few hundred agencies that never coordinated with each other, each with its own API, formats, vocabulary, and habits.
- The World Bank publishes annually and revises history without announcing it.
- FRED serves a daily series and a monthly one from the same endpoint, and the date strings look identical either way.
- The Treasury's monthly statement tables run on a trailing-twelve-months calendar, so a row's real calendar year depends on which month of the report you are reading.
- The EIA's generation data is a cube of fuel by state by sector, where most combinations are valid and some quietly are not.
None of these are defects. Each convention makes sense inside its agency. But it means that a chart comparing three sources is quietly running three different sets of assumptions, and any one of them handled wrong produces a chart that is confidently incorrect.
The work that makes it usable
Making the numbers chartable is mostly unglamorous engineering discipline. One normalized shape for every series, whatever the source looked like. Caching that stores the finest grain the source publishes and derives the rest, so a daily series can feed a monthly chart and a calendar heatmap from the same pull. Refresh rules that keep the last good copy when an upstream call fails, because a published chart that goes blank at 2 a.m. is worse than one that is a day stale. Corrections that propagate to every chart already built on the series, including the ones embedded in someone's article three months ago.
That work is the product. Statista built a nine-figure business substantially on it: professionals pay for numbers they can use in the next hour and cite with confidence, and the packaging is what makes that possible.
Where Kitegraph sits
We run that discipline across twenty-one sources today, growing steadily, with every series resolving to one shape that any of thirty-one chart types can read. The public sources are the start, not the identity: the same pipeline carries licensed and original data as the library grows. For the person making the chart, the effect is the same either way: pick the topic, and the data work has already happened.