Job Posting Observatory

Scope and methodology

1. 1.Sources: this site’s daily job sweep (MyCareersFuture endpoints and posting pages that are reachable without a login, plus the daily liveness log of the postings and market-pulse cards published here), joined with the first/last dates in the salary-disclosure ledger. Everything is public information; nothing is scraped from behind a login.
2. 2.Data layers: the automated daily revisit has run since 2026-07-21; every record before that is a backfilled head-and-tail observation from the salary ledger (employers whose rows are all backfilled are marked † in the table). For backfilled rows, days up are inferred from just two dates and observed-day counts are lower bounds; only the daily layer is confirmed day by day, and each posting only from the point it enters the daily panel.
3. 3.“Days up” runs from the posting date the job board itself printed (falling back to the date we first observed it) to the last date we observed it still up. It is a lower bound: our observation stopping does not mean the posting came down.
4. 4.“Multiple links, same role” counts cases where the same employer’s same normalised job title appeared under more than one posting link over time. That can be a re-listing, a double posting, or an agency posting running alongside a direct one. The data cannot tell these apart, so we count and do not interpret.
5. 5.“Over 45 days” is a descriptive threshold: the posting was observed up for more than 45 days. It implies nothing about the posting’s validity. The column currently holds 3 postings, of which 3 come from the backfilled layer (days up inferred from a first-seen and a last-seen date) and 0 have been confirmed day by day. These figures are recomputed from the ledger at every build; the line above the table prints the same breakdown.
6. 6.“Still under observation” means last observed within 3 days of the latest sweep.
7. 7.Coverage skew, stated up front: the automated daily sweep covers salary-disclosed postings in Singapore only; Hong Kong and UK observations come from manual filing at uneven frequency, and their last filing can predate the sweep date at the top of this page. “Postings observed” is what we saw, not what an employer actually posted, so employers should not be compared head-to-head on it, and markets even less so.