BLS User-Agent policy

It is BLS's stated condition for automated access: identify your client with a descriptive User-Agent carrying contact details, and avoid excessive request volume. The rule appears verbatim in the terms notes of all three BLS series in this slice: JOLTS (about 2,061 series, roughly 376,000 data rows), OEWS (413,528 rows in the May 2025 all-data file) and QCEW (county-level coverage each quarter).

• BLS Job Openings and Labor Turnover Survey (JOLTS) — "Automated retrieval must follow BLS usage policy: identify your client with a descriptive User-Agent including contact details and avoid excessive request volume."
• BLS Occupational Employment and Wage Statistics (OEWS) — "Scripted access must use an identifying User-Agent per BLS usage policy."
• BLS Quarterly Census of Employment and Wages (QCEW) — "Automated retrieval must follow BLS usage policy with an identifying User-Agent."

The underlying data stays US-government commercial delivery terms: no permission or fee is required, and BLS asks only that redistributed figures be credited with a release date. The policy governs how you retrieve, not whether you may use.

Because a blocked crawler is an outage. Scripts that hit federal endpoints anonymously or at high volume risk throttling, and unlike commercial APIs there is no paid tier to buy reliability back.

The economics also differ from most commercial sources: these three series are free and commercial delivery terms, so the only scarce resource is your access path. QCEW alone publishes county x industry x quarter employment for every US county - its NAICS 5613 Employment Services slice runs 3,623 rows in Q1 2025 alone - which makes it exactly the kind of table teams script against. OEWS ships 413,528 rows in the May 2025 all-data file and JOLTS about 2,061 series with roughly 376,000 data rows, so unthrottled re-downloads waste both sides' bandwidth.

Compliance cost is near zero - one header string - while the downside of ignoring the policy is losing scheduled refreshes on the wage and turnover benchmarks many HR products are built on.

1. Send an identifiable header. Include a product name and a real contact email in the User-Agent string, as JOLTS, OEWS and QCEW terms notes each specify.
2. Throttle to reasonable volume. Space requests; 'avoid excessive request volume' is the published standard.
3. Cache aggressively. JOLTS updates monthly and OEWS annually, so re-downloading daily buys nothing.

Where to see it in context: human-resource-employment-services data.

Entries adjacent to the BLS User-Agent policy:

• membership-gated-data — the contrasting access model where detail sits behind association membership
• human-resource-employment-services data — the industry slice whose 3 BLS series (JOLTS, OEWS, QCEW) cite this policy