How to Scrape BLS Labor Statistics

The Bureau of Labor Statistics publishes thousands of useful series, but a raw API response still needs period parsing, numeric conversion, limits, and provenance before it belongs in a production dataset. The BLS Labor Statistics Scraper handles that connector layer while keeping the official series ID beside every observation.

Start with known series IDs

Use CUUR0000SA0 for headline CPI, LNS14000000 for the civilian unemployment rate, CES0000000001 for total nonfarm employment, or another code from the BLS series catalog.

{"seriesIds":["CUUR0000SA0","LNS14000000"],"startYear":2024,"endYear":2026,"maxResultsPerSeries":36}

Each saved row represents one period in one series. The dataset includes year, BLS period code, readable period name, numeric value, footnotes, latest-record flag, and an official source link. Keep series_id as the primary dimension; friendly titles are supplied for common presets but should not replace the canonical code.

The keyless tier supports a ten-year window and up to 25 series in one request. Split longer historical research into bounded Tasks instead of silently expecting the public endpoint to accept a larger query. Annual averages can be requested where BLS publishes them.

For scheduled monitoring, store collection time downstream and upsert on series ID, year, and period. BLS values can be revised, so append-only ingestion can create conflicting observations. A revision-aware table preserves the latest value while keeping earlier snapshots for audit. Empty series should be treated as a validation signal, not as zero economic activity.