Check the data before you pay.
Thirteen CSV samples cut from the same build the files are sold from. They are real rows chosen at random (seeded, so they do not change between builds), names unmasked only where the legal form is not a person's name, and with no contact fields: the Brazil samples withhold phone and e-mail, the unified schema never had them.
Download
Reading the samples
Every file starts with one comment line naming the source and the release, then a header row. Dates are ISO (YYYY-MM-DD), booleans are True/False, lists are joined with ';'. Encoding is UTF-8 with a comma separator; Excel on Windows may need the import wizard for accented names.
The unified schema has 44 columns and is the same for every country, so a column that the registry does not publish is simply empty for that country (address in Ecuador, website outside Brazil and Mexico, size in Argentina). The methodology page lists what each registry publishes; each country page shows coverage per column.
Need a sample of a specific cut (one region, one division, one size class) before buying? Ask: a 25-row custom sample is free.
Happy with what you see?
Packages ship in one business day with CSV, Parquet and a data dictionary.