Salary Data from Job Postings - Greenhouse, Lever, Ashby
Pricing
from $5.00 / 1,000 pay records
Reads open postings on the public Greenhouse, Ashby and Lever job boards you name, extracts employer-published pay ranges, normalizes them (min/max/currency/interval, role family, seniority) and rolls them up into median, p25 and p75 salary benchmarks.
Company slugs on Greenhouse. The slug is the last path segment of the company's board URL: job-boards.greenhouse.io/anthropic -> "anthropic". Leave as-is to use the built-in seed list. A full URL is accepted and reduced to its slug.
[ "anthropic", "databricks", "datadog", "gitlab", "cloudflare"]Company slugs on Ashby. The slug is the first path segment of jobs.ashbyhq.com/openai -> "openai". Ashby publishes a typed compensation field, so coverage is usually higher than Greenhouse.
[ "openai", "ramp", "sierra", "harvey", "notion"]Company slugs on Lever, e.g. jobs.lever.co/leverdemo -> "leverdemo". Empty by default: many companies that once used Lever now return 404, and Lever postings carry pay ranges less often than the other two.
[]On by default. Turn it off to also receive postings with no published pay, which is useful for measuring what share of a company's roles disclose. Postings without a range are never charged for.
Hard cap on dataset records, and therefore on cost. The run stops as soon as it is reached. Range 1-200000.
Caps how many postings are read from each company, so one very large board cannot use up the whole record budget. 0 means no per-board limit.
Keep only these normalized role families. Leave empty for all. The family is derived from the job title and department.
[]Keep only these seniority levels, derived from title tokens. Leave empty for all. "unspecified" covers titles with no level word, which are usually mid-level.
[]Case-insensitive substring match against the posting's raw location and its normalized bucket, e.g. "New York", "CA", "Remote". Leave empty for all locations.
ISO codes such as USD, GBP, EUR. Leave empty for all. Note that this actor never converts between currencies.
[]Drop postings whose annualized maximum falls below this figure, in the posting's own currency. 0 disables the filter.
Writes median/p25/p75 rows grouped by role family x seniority x location x currency to the named "benchmarks" dataset and to the BENCHMARKS key-value record.
A group needs at least this many postings before a benchmark row is emitted. Below about 3 the percentiles are not meaningful.
These are public JSON endpoints with generous limits, but 5 is polite and plenty fast - each request returns an entire board.