Radar Labs / salary dataset / freshness

How old are the postings that are still "open"?

8,041 live postings read on 2026-10-24 from 259 company boards, measured from each board's first-published field where it has one.

Median age

51d

p10 8d · p90 233d

Older than 30d

65.9%

5,296 of 8,041

Older than 90d

29.6%

2,383 of 8,041

Older than 180d

14.0%

1,125 of 8,041

Older than 365d

4.7%

380 of 8,041

By board — and this is where it gets interesting

Board	Postings	Median age (d)	p90 (d)	Oldest (d)	>180d	Date field
greenhouse	5,134	52	233	2628	14.3%	first_published
ashby	2,743	45	213	1325	12.4%	createdAt
lever	164	79	514	2230	30.5%	publishedAt

The date field column is the point. A scraper that reads updated_at on Greenhouse gets "last edited by a recruiter", not "posted": a requisition from 2024 shows up as a day old the moment someone fixes a typo. Our first pass did exactly that and reported 0% stale Greenhouse postings. Reading first_published instead changes the answer to 14.3%.

Companies with the oldest open requisitions

Company	Postings	Median age (d)	Oldest (d)	>180d
secureframe	15	442	2,000	93.3%
zilliz	7	430	1,248	71.4%
character	13	310	1,004	69.2%
delfina	27	253	515	55.6%
wrike	6	205	1,284	50.0%
cognition	9	138	366	44.4%
doxel	12	137	318	41.7%
circleci	12	84	444	41.7%
typeface	5	53	582	40.0%
databricks	473	115	2,536	38.3%
figma	79	94	634	36.7%
scaleai	116	107	1,229	36.2%
stockx	14	59	262	35.7%
hex	37	58	906	35.1%
bloomreach	23	85	2,628	34.8%
modal	42	86	731	33.3%
encord	9	124	207	33.3%
honehealth	6	172	554	33.3%
togetherai	41	114	1,012	31.7%
entrata	19	79	1,774	31.6%
melio	13	86	306	30.8%
labelbox	7	137	200	28.6%
cohere	78	53	722	28.2%
rain	37	85	288	27.0%
zoominfo	15	44	570	26.7%

Only companies with at least five dated postings. A stale listing is not proof of a fake job — plenty of teams keep a rolling req open for a role they hire into continuously.

What to do with this

If you are filtering scraped jobs by date, sort by first published, not by last modified, and watch for boards that only expose the latter. If you are the one reading the list rather than building it: an "open" role older than six months is usually a pipeline-building req, and it will rarely move fast.

Raw numbers: freshness.json. The postings themselves, with dates and pay ranges: the dataset (CC BY 4.0).