2026 Study · Second wave
More than 1 in 5 large-employer job postings fails state pay-transparency law
We collected 300 recent job postings from 30 large U.S. employers — the same 20 panel employers as our first wave, plus 10 fresh employers — and graded each one against the pay-transparency posting requirements of the state it was hiring in, using the same frozen requirement dataset and the same method as wave 1. Of the 84 postings that could be determinately graded, 19 failed at least one requirement — a 22.6% fail rate (Wilson 95% confidence interval 15.0–32.6%). The most common miss: the pay range itself.
Seven of the 30 employers had at least one failing posting — all seven are panel employers. No fresh employer had a failing posting in this wave.
Key findings
1. 22.6% of gradable postings fail.
Of the 84 postings hiring in states with in-force pay-transparency posting requirements, 19 were missing at least one required element — a pay range or an application deadline. That's more than 1 in 5. The headline is the wave-2 cross-section: 300 postings, 84 graded, 65 passing, 33 indeterminate, 183 ungradable.
2. The pay range is the most-missed requirement.
Fourteen of the 19 failures (74%) were missing the pay range itself — the central disclosure these laws were written to require. Application deadlines accounted for 7 failures (all in Colorado). No failure in this wave was on benefits information alone, and no failing row was missing benefits information at all. A row can fail more than one requirement.
3. Panel vs. fresh: the failures are all panel.
Panel employers only: 35.8% fail (19 of 53 determinately graded). Fresh employers only: 0 fails (0 of 31). The panel comparison with wave 1 (17.1%, 13 of 76) is directionally higher, but denominators and composition differ and the confidence intervals overlap — this page does not claim a proven rise in non-compliance.
4. Colorado is the outlier.
All 7 Colorado determinate rows fail, all on application deadline (5 also fail pay). Colorado is the only jurisdiction in the dataset requiring an application deadline, and “applications accepted on an ongoing basis” does not count as a specific deadline date under the study's method. That rule is disclosed, not smoothed over.
5. 216 postings couldn't be graded — and we didn't count them as passes.
Most of the sample was excluded from grading: postings in states with no applicable posting requirement, postings whose location couldn't be mapped to a jurisdiction, summary rows where a missing benefits or deadline element couldn't be determined, and 2 lease-holder listings (a non-standard employment category). Excluded postings are reported here — they are not treated as compliant. Six panel employers contribute zero determinate rows in wave 2 — that is not a clean bill of health for them.
State breakdown
| State | Determinate | Pass | Fail |
|---|---|---|---|
| Washington | 16 | 13 | 3 |
| California | 24 | 22 | 2 |
| New York | 14 | 11 | 3 |
| Illinois | 12 | 12 | 0 |
| Colorado | 7 | 0 | 7 |
| Virginia | 7 | 5 | 2 |
| Minnesota | 6 | 5 | 1 |
| Massachusetts | 3 | 3 | 0 |
| New Jersey | 2 | 1 | 1 |
| District of Columbia | 2 | 2 | 0 |
| Maryland | 1 | 1 | 0 |
| Connecticut | 1 | 1 | 0 |
Raw counts only — no state reaches 30 determinate rows, so no state rate is quoted. State rows sum above the determinate total because multi-state rows count once per applicable state. Indeterminate rows by state: WA 7, CA 2, NY 1, IL 6, CO 6, MA 1, MD 1.
What this means for employers
Your posting process has a gap, not a conspiracy. The pattern — missing pay ranges across a quarter of large employers in the panel, spanning industries — looks like inconsistent process, not deliberate defiance. If pay-range disclosure isn't a hard gate in the drafting chain, it falls off. It did, in more than 1 of 5 gradable cases.
The risk is compounding. Pay-transparency posting requirements now cover 15 U.S. jurisdictions, and more states are adding them. The consequences are no longer theoretical: more than 250 Washington employers have faced class actions over deficient postings since that state's law took effect — including a $1.1–$1.65M settlement (alleged violations, not an adjudicated finding) — and New York City can fine up to $125,000 per uncured violation.
The fix is cheap. A deterministic pre-posting check — pay range present, benefits described, deadline stated — takes seconds per posting and catches exactly the failures this study found. We built one: it's free, it runs in your browser, no signup.
Non-compliant postings don't just risk fines — they lose candidates. Two separate 2026 surveys (Monster, iHire) find six in ten job seekers won't apply to — or are less likely to apply to — a posting without a salary range. A missing pay range is a smaller applicant pool before it is ever a legal problem.
Grade your postings in bulk → Check a single posting →
Methodology
The sample. 300 job postings from 30 large employers (1,000+ employees each), 10 per employer: the same 20 panel employers as wave 1, plus 10 fresh employers chosen for industry balance (zero retail — retail was already over-represented in the panel). For each employer: the 10 most recently posted roles on their public career page as of the collection date, excluding internships, contractor/gig roles, and internal-only transfers.
Collection window. Postings were collected overnight October 27–28, 2026, and graded against the requirement dataset frozen as of October 27, 2026. The dataset is unchanged from wave 1: 15 posting jurisdictions in force (CA, CO, CT, DC, HI, IL, MA, MD, ME, MN, NJ, NY, VA, VT, WA). No in-force law changes occurred between wave 1 and wave 2. Upcoming items are disclosed, not graded: Delaware (September 2027), Columbus OH (January 2027), Washington cure-window end (July 27, 2027). NV and RI are on-request states and ungradable, as in wave 1. Wave-1 rows were never re-graded.
The grader. Each posting was graded with a deterministic heuristic scan against our verified pay-transparency requirement dataset — the same engine behind our public bulk audit tool. It checks for the presence of a pay range, benefits description, and application deadline wherever each is required. Deterministic means reproducible: anyone re-running the same postings gets the same results. No AI involved, no judgment calls.
What counts as a fail. A posting fails if any required element is missing for its jurisdiction in the captured posting text. A posting passes only if all applicable requirements are detected. Full-text rows: a missing applicable element counts as a fail. Summary rows: pay determinations are reliable, but a missing benefits or deadline element on a summary row counts as indeterminate for that element, not a fail. Indeterminate rows are never counted as fails. A single starting rate is not a pay range; an ongoing-basis note is not a specific deadline date.
Dual grading. An automated pass plus human review of every failing row and 20% of passing rows. The review flagged one employer's six rows as artifact-risk — pay and benefits absent from every fetched full text — and a live re-check of those detail pages found a populated pay and benefits section the text fetch had missed. Those six rows were corrected to passes before publication, and the correction is disclosed here rather than hidden. A control posting in a non-covered state genuinely had no pay section, confirming the miss was in the fetch, not invented.
Top-ups. The pre-registered ladder toward 500 collected postings was not dispatched. Determinate yield in wave 2 is 28% (84 of 300), not the 38% assumed from wave 1, so 500 collected would add only about 56 determinate rows (~140 total), still below the 150 threshold, let alone a 190 target. This wave is published at 300 with wider confidence intervals and no subgroup claims, per the protocol's below-150 rule.
Live-status honesty. Grading is on captured postings, not on live status at publication. Where a live check found a captured row unavailable or not found, grading used the original captured text: one employer's captured top 10 does not match its board's current top 10 (board turnover — the captured set stands); two other employers' February/March rows were not confirmed open; another employer's original rows were not recovered live, and no $45/hour figure was found in its live listings (drivers shown at $23/hour, warehouse at $21/hour); and two rows from a fourth employer returned 404 on re-check, with linked PDF figures treated as supplements only, not posting-text pay.
Anonymization. Employers are not named. The grader is a heuristic screen, not a legal determination, and naming an employer as “non-compliant” on that basis would overclaim what the data supports. The aggregate story is the finding.
Honest limits
- This is a sample of 300 postings from 30 large employers — not a census of U.S. job postings. It describes the sample, nothing more.
- The grader is a fast heuristic scan: pattern matching for pay figures, benefits language, and deadlines. It can miss things and misread things — as the fetch miss corrected above shows. It is not legal advice and not a legal determination of compliance.
- “Checked, never certified”: a passing grade doesn't mean a posting is legally compliant, and a failing grade doesn't mean the employer broke the law.
- State assignment follows the location listed on the posting — an approximation, especially for remote and multi-state roles.
- Postings and laws both change. These are results for the collection window stated above.
- This page is not the remote-posting study. That study's collection is incomplete and its merge and grading have not run; nothing in this page's numbers comes from it.
This restraint is deliberate. A study that overclaims gets one news cycle and zero repeat citations; a study that's honest about its limits gets cited every year.
Press
For press inquiries, the anonymized dataset, or a quote: jobdescriptiontool@gmail.com
“This isn't employers defying the law. It's posting processes that were never rebuilt for a multi-state compliance reality — and the fix is a check that takes seconds.”— Ryan Stanford, founder, jobdescriptiontool.com
jobdescriptiontool.com is a free toolkit for compliant hiring: a pay-transparency checker covering 15 U.S. jurisdictions, an AI job description generator, posting audits, and offer tools. Every state requirement in its dataset is sourced from official labor-department publications.