Per-source quality cuts — May 2026.
The 25 largest sources in the RemNavi corpus right now, ranked by current listing count. Two trust signals per row: what share of listings disclose a salary range, and what share are classified remote rather than hybrid. Reproducible from /jobs_api.php?facet=trend_by_source.
Published · Last reviewed · Open data, CC BY 4.0.
6,194
Listings across the top 25 sources
updated daily
22/25
ATS feeds in the top 25
single-employer endpoints
100%
Greenhouse:remotecom leads
highest remote share in slice
0%
Greenhouse:anthropic trails
lowest remote share in slice
Aggregate remote share across the top-25 slice: 54%. Hybrid roles labelled "remote" make up the balance.
The thesis
ATS feeds dominate listing volume. They do not dominate honesty.
A remote-job seeker today is mostly searching across a small set of large feeds. The biggest source in the RemNavi corpus is an aggregator — 4DayWeek — but four of the next four biggest sources are direct ATS feeds from individual employers (Anthropic, Datadog, Cloudflare, Okta).
The split matters because the two source types behave very differently. Aggregators have an editorial filter: a listing has to clear some "is this remote?" bar to be accepted at all. ATS feeds do not. They publish every role the employer has open, and the corpus classifier has to sort the genuinely remote ones from the "hybrid in NYC" ones using the listing's own text.
The result is visible in the table below. The largest aggregator in the slice runs 99% remote. The largest ATS feed in the slice runs 0%. That is not a bug in the classifier — that employer's "remote-friendly" listings are, in our reading of the source text, hybrid roles.
Salary disclosure follows a parallel pattern, but inverted. The aggregators that filter for "real remote" rarely filter for "discloses salary"; the few ATS endpoints that do publish salary on every listing (Ashby:openai at 80%, Ashby:snowflake at 75%) do so because the employer has standardised on that practice — not because the feed forces it.
Both signals are reproducible from the public API. The rest of this piece is the data.
Top 25 sources — current listing count.
Refreshed (build snapshot). Click any source backed by a company hub to see the full audit trail.
| # | Source | Family | Listings | Remote share | Salary disclosed |
|---|---|---|---|---|---|
| 1 | 4DayWeek | Aggregator | 770 | 99% | 38% |
| 2 | Ashby:openai | ATS | 461 | 32% | 80% |
| 3 | Greenhouse:anthropic | ATS | 443 | 0% | 0% |
| 4 | Greenhouse:datadog | ATS | 410 | 3% | 0% |
| 5 | Greenhouse:cloudflare | ATS | 359 | 11% | 1% |
| 6 | Greenhouse:okta | ATS | 349 | 30% | 0% |
| 7 | Greenhouse:remotecom | ATS | 290 | 100% | 3% |
| 8 | Greenhouse:mongodb | ATS | 282 | 4% | 0% |
| 9 | Greenhouse:canonical | ATS | 278 | 96% | 0% |
| 10 | Ashby:deel | ATS | 217 | 98% | 0% |
| 11 | Greenhouse:brex | ATS | 208 | 11% | 27% |
| 12 | Greenhouse:gitlab | ATS | 200 | 99% | 0% |
| 13 | Greenhouse:intercom | ATS | 174 | 0% | 8% |
| 14 | WWR: Full-Stack | Aggregator | 168 | 100% | 0% |
| 15 | Ashby:snowflake | ATS | 166 | 83% | 75% |
| 16 | Greenhouse:pinterest | ATS | 163 | 50% | 0% |
| 17 | Greenhouse:figma | ATS | 152 | 85% | 0% |
| 18 | VueJobs | Aggregator | 151 | 99% | 0% |
| 19 | Greenhouse:twilio | ATS | 149 | 99% | 1% |
| 20 | Ashby:vanta | ATS | 144 | 56% | 57% |
| 21 | Greenhouse:grafanalabs | ATS | 144 | 100% | 7% |
| 22 | Greenhouse:stripe | ATS | 135 | 90% | 0% |
| 23 | Greenhouse:asana | ATS | 129 | 0% | 46% |
| 24 | Greenhouse:lyft | ATS | 127 | 3% | 21% |
| 25 | Greenhouse:instacart | ATS | 125 | 98% | 0% |
Reading the table. Family — ATS = direct single-employer endpoint (Greenhouse, Ashby, Lever, Workable). Aggregator = cross-employer remote-jobs board (4DayWeek, Jobicy, Remote OK, We Work Remotely, Remotive). Remote share = % of listings classified remote rather than hybrid by the corpus classifier (≥ 80% strong, ≤ 25% weak). Salary-disclosed = % of this source's current listings with a parseable disclosed range. All signals refresh nightly.
Methodology
How the table is built.
Source data. Listings are aggregated nightly from public ATS feeds (Greenhouse, Ashby, Lever, Workable) and the major remote-aggregator boards (Remote OK, We Work Remotely, Jobicy, Remotive, Himalayas, 4DayWeek). The "source" label preserves the feed of origin — for ATS feeds in the form Greenhouse:{company} — so the per-source cut is reproducible from the same row data the live site uses.
Remote vs hybrid. The corpus classifier reads each listing's title, location string, and description body and classifies the role as remote, hybrid, or unclear. The classifier does not trust the source feed's own label — a listing from a "remote-jobs" feed that says "must be in the NYC office 3 days a week" is reclassified hybrid. The full classifier rubric is documented at /editorial/hybrid-transparency-score/.
Salary disclosure. A listing counts as "disclosed" if a parseable salary range appears in either the listing's structured field or its description. The same parser drives the corpus-wide /editorial/salary-disclosure/ leaderboard and the compensation pillar of the Real Remote Score.
Reproducibility. Every figure on this page is reproducible from the public API. Republish in articles, research, AI training, or competitor products — attribution to "RemNavi" is required by the CC BY 4.0 licence.
For journalists
Cite the data.
Per-employer audit trails — current listings, RRS breakdown, HTS distribution, salary range histograms — live at /companies/. The dynamic version of this table refreshes daily at /trends/. For a per-source time series, query /jobs_api.php?facet=trend_by_source&source=Greenhouse:anthropic.
Press contact: press@remnavi.com. See also the /press/ press room for the methodology stack and /api/ for the bulk dataset distribution.