Which tech roles are growing in Europe

Method and limits

Reading the columns. Roles are distinct (one company, one title), so a company that posts the same role in nine cities counts once; the board itself counts postings. New, 30d is the share of a family's postings posted in the last 30 days. Est. comp is the median of the board's estimated total compensation for on-site and hybrid roles at tiered companies, with at least 20 priced roles (the methodology explains the estimate). Competition is member applications per open role over 90 days, indexed to the engineering average of 100, a ratio never a count, blank under the sample floors. Change compares today with a weekly snapshot of the same count on the same companies, the one exact trend here. 12w share compares a family's share of roles twelve weeks ago with today on the same companies, floored at 100 roles; it is marked approximate while it is read back from our records, and switches to the weekly snapshots on its own once two of them span the window.

The source is the curated board behind /jobs: on-site, hybrid and remote roles at tiered tech companies hiring in Europe, ingested from company career pages and job APIs and filtered to Europe. It is a curated panel, not the whole market: a family's count moves when we add a company to the panel, and the change column reads that as growth. Treat shares and changes as the reading of one large, consistent sample.

Families are assigned from the title alone by a fixed dictionary: the function noun wins over the domain qualifier (an "Account Executive, Security" is sales, a "Security Engineer" is security, a "Product Manager, Security" is product management), an engineering title is never filed under a business family, and a title with a specific domain (forward deployed, AI, embedded, SRE) lands there before generic software engineering. Titles no rule reads, mostly non-English operations roles, are counted as other and shown as such. Changing a rule restarts that family's change column.

The approximate 12-week column is rebuilt from the rows we hold, live ones and our deletion log, counting at each past Monday the roles posted by then and not yet closed. Three things are isolated: only companies our pipeline already held at the start of the window count, so a company we added later never reads as growth; deletions that retired a duplicate of a row that lives on are not closures; and roles are distinct, so a company that posts one role in nine cities counts once. Two things cannot be isolated, which is why the column shows shares and never counts: our own clean-up runs move the total by a quarter in a week while the big families' shares move by a point or less, and for most on-site roles the posting date we hold is the day we first saw the role, not the day it opened. Once two weekly snapshots span the window the column switches to them on its own and loses the approximate mark. The exact change column already compares the same companies across its two snapshots, so our own additions to the board never read as a market move there.

The competition index is a ratio over paying members' apply clicks, never a count of them, and is only shown above sample floors. It describes this site's members, who skew senior and European, not every applicant in the market. The pay median is the board's own estimate for software and non-software roles at tiered companies, not posted salaries; remote roles carry no estimate and are excluded from it.

Related: the Compensation Report for what engineers report earning, where to work in Europe for the same board by country, and the board itself, where the Category and Workplace filters cut the live roles.