How Much a Job Posting Tells You Is Decided by the Board, Not the Employer

Bars at 5,000 are postings sitting on our own storage cap, so those feeds carry at least that much and possibly more. The short end is the honest end: nothing in our pipeline can make a description shorter than the feed delivered it.

Every employer with at least 40 active postings on each of two or more boards, 24 of them in the corpus. The last two rows are counter-examples and they matter: for 14 of the 18 employers where a median could be computed on both sides the two boards agree to within 10%, so this is a property of specific feeds rather than a universal law. A median printed as 5,000+ is sitting on our storage cap and could be longer, which makes every ratio here a floor.

Source: Seeker job corpus, 257,534 active job postings, snapshot 2026-08-16. seekerscore.com/insights/how-much-a-posting-tells-you-2026

The finding

9.5% of live postings carry under 500 characters of description, which is roughly a tweet and a half to decide a job application on. That is not spread evenly. It is concentrated in a few feeds: oracle_cloud runs a median of 123 characters with 74% under 300, reed puts 100% of its postings under 500, jooble the same, and adzuna delivers descriptions hard-capped at exactly 500 characters on every single row. Those four feeds account for 36,429 active postings between them.

Why this is the channel and not the employer

The test that separates the two is holding the employer fixed and varying the board, which is possible for the 24 companies posting at least 40 active roles on two or more feeds. JPMorgan Chase is the clearest case in the corpus: 12,046 postings through oracle_cloud at a median of 120 characters with 98% under 500, and 41 postings through TheMuse where none are under 500 and the median sits on our cap. Same bank, same weeks, 41.7x difference in how much you are told. Apple shows the same shape at 5.3x through its own careers feed against TheMuse, and Included Health at 6.3x between lever and himalayas. Nothing about the employer changed. What changed is which pipe the posting came down.

The counter-examples, which are half the result

This is not a universal law and saying so would be the easy overreach. For 14 of the 18 employers where we can compute a median on both sides, the boards agree: Stripe reads the same through greenhouse, findwork and himalayas; Twilio, GitLab, Grafana Labs, Reddit, Salesforce, Sanofi, Zscaler, Smartsheet, Mercury, Hightouch, Abnormal Security, Allstate and Nationwide all land within 10% of themselves across feeds. So the honest claim is narrow: a specific set of feeds truncates, and when you hit one of them the employer's own writing is not what you are reading. It is not that job boards in general show you less.

What this does to a job search

Two practical consequences. First, a thin posting is not evidence of a lazy or low-quality employer, so do not filter on it. The same role may exist in full somewhere else, and for the largest offenders it demonstrably does. Second, and less comfortable, a 120-character description cannot state a salary, a seniority level, or a years requirement, which means postings arriving through those feeds are structurally missing the fields every filter depends on. If a search tool tells you a role has no stated experience requirement, the honest reading is often that the feed never carried one.

How we measured it (and what we did not claim)

Population: a 20% systematic sample of the 257,534 active postings for the per-source figures, 52,300 of which carry a description, plus a full-corpus pass restricted to multi-board employers for the control table. Length is the character count of the description text as stored. The critical limit, stated up front because it decides which half of this page is trustworthy: we cap stored descriptions at 5,000 characters at write time, and 49.6% of active postings sit on that cap. That makes the LONG end of every distribution censored and unknowable, which is why this page reports no average length, no long-tail figure and no claim about which board is most thorough. The SHORT end is unaffected, because nothing in our pipeline can make a description shorter than the feed delivered it, so every figure here is about the bottom of the distribution only. Ratios involving a capped median are floors rather than estimates. We are NOT claiming these employers write short job descriptions, and we are NOT claiming the truncation happens at the employer. For oracle_cloud, reed, jooble and adzuna the text arrives already short, and we cannot see whether the board truncated it or the employer's own posting system did.

Frequently asked questions

Related Data

Guides

• Job Search Statistics →
• Hidden Job Markets →
• Resume Keywords That Matter →