How complete is our own sample data: 83.5% of fields filled across 24 files

Every actor has a free sample CSV. We computed, for every column in every sample, the share of rows where the value is non-empty. That is 552 columns over 216 rows in total (4 to 12 rows per file).

• Mean coverage across all columns: 83.5%; the median file averages 92.1%.
• 365 of 552 columns (66%) are filled in every row. 8 columns are empty in every row, spread over 5 files.
• Least complete files: app-store-reviews (38.6%), developer-signals (63%), website-contact-extractor (70.4%). Most complete: building-permits-leads (100%), new-business-filings (100%), sitemap-link-inventory (100%).
• Contact columns (anything named email or phone): 19 columns, mean coverage 57.4%, and only 7 are filled in every row.

Public sources do not publish every field for every record, and a contact detail is the field most often missing. Judge a sample by the columns you need, not the average. The data quality pages show the same coverage table for each source.

Limits

• Samples of 4 to 12 rows. A column that is 100% filled in 5 rows can be half empty in 5,000.
• "Empty" means blank, None, [] or {}. A column that holds a value like "unknown" counts as filled.
• The samples have personal contact data masked, which does not change whether a value is present.

Data and method

Computed by a5_coverage.py over the CSVs in apify-actor-samples (pass the folder path as the first argument). Published 2026-10-13. We build and maintain these actors.