How We Measure the Job Market

This research is built on 483,233 job postings we collected and deduplicated to 204,223 unique jobs from 30,910 companies. Salary figures are platform-estimated posted ranges — not employer disclosure — and the corpus is tech & professional, not all US jobs.

Everything you need to evaluate, trust, and cite SoviaJobs Job Market Research.

SoviaJobs ResearchData through June 2026

Corpus & window

SoviaJobs continuously collects job postings as candidates browse and apply across the US market. For this research we used a snapshot of 483,233 collected rows spanning 2026-03-20 to 2026-06-09 (~81 days).

Source coverage

Postings arrive from major job boards and directly from companies’ own career sites and ATS instances (Workday, Ashby, Greenhouse, Lever, and dozens more). Direct ATS and company sites account for roughly 37.8% of collected postings — this is not just an aggregator scrape.

Deduplication

The same job is often collected many times. We deduplicate on the platform’s jobPostingId, collapsing 483,233 raw rows down to 204,223 unique postings before any aggregation. Reposting behavior itself is reported separately in the Job Reposting Report.

Salary honesty

Our median posted midpoint is $144,000 (p25 $115,000 / p75 $177,500). Critically, these are platform-estimated posted market ranges — the pay band shown to candidates (posted or estimated) — not a measure of employer salary disclosure. Only ~9% of description texts contain a literal dollar figure, so we never present these figures as a “salary transparency” rate.

Tech skew & blind spots

SoviaJobs is a tech & professional job-application platform. The corpus over-weights software, data, and product roles (the single most common required skill is Python). We label findings as the “tech & professional” job market, never “all US jobs.” We also cannot see: applications-per-posting, time-to-fill, who got hired, or full description text (it is truncated to ~2K characters), so description-level shares are lower bounds.

Benchmarks

Our median posted midpoint of $144,000 cross-checks against an independent v2 pass at $142K — agreement to the rounding level. Where trends are directionally comparable we frame them against BLS JOLTS (openings/hires) and the Indeed Hiring Lab, while being clear those measure a broader, all-industry universe than our tech & professional corpus.

License & citation

This research is free to use with citation and a link to soviajobs.com/market-research (CC BY 4.0). Every report has a one-click “copy citation” button and an embed option. Suggested citation:

Source: SoviaJobs Job Market Research, June 2026, soviajobs.com/market-research

Data downloads

Raw, flattened CSVs for each dataset (license header included):

Prefer markdown? Every report has a plain-text mirror at /market-research/reports/<slug>.md.

Press & embeds

For data requests, custom cuts, or interviews, contact press@soviajobs.com. To embed a chart, use the “Embed” button on any report — it emits an iframe snippet with the attribution link baked in.

Methodology FAQ

How many job postings is this research based on?
We collected 483,233 raw postings and deduplicated them to 204,223 unique jobs from 30,910 companies, over the window 2026-03-20 to 2026-06-09.
Are these salaries what employers actually disclose?
No. The salary figures are platform-estimated posted market ranges — the pay band a candidate sees on the listing, whether the employer posted it or it was estimated. Only about 9% of description texts contain a literal dollar figure, so this is not a measure of salary-disclosure compliance.
Does this represent all US jobs?
No. SoviaJobs is a tech and professional job-application platform, so the corpus is tech- and professional-skewed (the single most common skill is Python). Read every figure as the tech & professional job market, not all US jobs.
Can I cite or republish these numbers?
Yes — the data is free to use with citation and a link back to soviajobs.com/market-research. Each report has a 'copy citation' button, downloadable CSVs, and an embed option.
How often is the data refreshed?
The snapshots are regenerated monthly from the live dataset. The 'Data through' stamp on every page tells you the current window.