Web to Table

Request type

Research and obscure data

Pull scattered public data out of PDFs, government pages, registries, and archives into one clean table.

Planned

Planned. Sources are checked one by one before you pay.

Who asks for this

Statisticians, academics, journalists, and analysts whose data exists but lives in fourteen different places.

Say it like this

Collect the annual rainfall tables published by these 14 regional water authorities into one table, with the source and year for every row.

What you supply

  • The sources, or a description of them
  • The fields you need per record
  • The years or date range

What you get back

Your table · 4 rows shownIllustrative
regiontext · collectedyearnumber · collectedrainfall_mmnumber · collectedsource_formattext · derivedsource_urlurl · collected
North Basin2,0241,120pdfnorthbasin.example/annual-2024.pdf
Coastal2,0241,580htmlcoastalwater.example/stats
Highlands2,024pdfhighlands.example/report24.pdf
Valley2,024640htmlvalleywater.example/data
The Highlands report did not include 2024 rainfall, so the cell stays empty.

Column labels: collected comes from the page, derived is calculated from other columns at no extra cost, extra work needs another request per row and raises the estimate.

Workflow

How the agent handles it.

  1. 1

    You list the sources or describe where the data lives.

  2. 2

    We check each source and tell you which fields exist and in what format.

  3. 3

    You approve the estimate.

  4. 4

    We extract, normalize units, and attach a source URL to every row.

  5. 5

    You export the table with its sources, ready to cite.

Cost and limits

What moves the price, and where the answer can be no.

Cost drivers

  • Number of sources
  • PDFs and scanned documents, which cost more to read than web pages
  • Whether the fields need reconciling across formats

Where we say no first

  • A source may not publish every field for every year.
  • Scanned documents can be misread. We flag low-confidence values.

Early access

Want research and obscure data?

Describe your sources, the columns you want, and roughly how many rows. A person replies with possible or not, and roughly what it would cost.