Directive (EU) 2023/970 has been Italian law since 7 June 2026, transposed by D.Lgs. 96/2026. Among other things it means a candidate is entitled to pay information before the interview, not after it.

Plenty of people are now selling opinions about how much that has changed job ads. We hold 10,371 open postings, so we counted ours instead. Everything below is one query against the live board on 25 August 2026, and the limits are at the bottom because they matter more than the headline.

What we found

Across the whole board — 10,371 open roles

Publish a salary band: 1,166 — 11.2%

Italian postings — 877 open roles

Publish a salary band: 233 — 26.6%
Name the collective agreement (CCNL): 82 — 9.4%

Italian ads on our board disclose pay at more than twice the rate of the board overall. That is the good news, and it is genuinely good.

The CCNL figure is the one worth sitting with. Article 5 of the Directive is about what a candidate must be told before applying, and the applicable collective agreement is part of how pay is actually determined in Italy. Fewer than one Italian posting in ten names it.

Who publishes pay, and who does not

By field, across every country, among open roles:

Media: 22%
Technology: 20%
Healthcare: 14%
Design: 14%
Architecture: 12%
Science: 12%
Engineering: 10%
Business: 9%
Law: 8%
Trades: 7%

And the sharpest split we found is not a sector at all:

Remote roles: 26% publish pay
Everything else: 9%

Roles advertised as remote disclose pay at nearly three times the rate of on-site ones. We would guess that hiring across borders forces the question earlier — you cannot let a candidate infer the number from the city. But that is a guess, and we have not tested it.

What we will not tell you, and why

Whether the law moved the needle. We cannot say. Our board holds live postings, so a role first posted in April and still open in August is not a typical April role — it is a role nobody has filled in four months. Any before-and-after comparison across 7 June is measuring how long jobs stay open at least as much as it measures disclosure. The monthly Italian samples are also small: nine postings in April, thirty-two in May. A percentage of nine is not a finding.

If someone shows you a chart of Italian pay disclosure rising or falling week by week, ask them how they handled that, because it is the whole problem.

That this describes the Italian labour market. It does not. It describes the employers who publish through the applicant-tracking systems we read, which skews corporate, and skews towards the fields we carry most: engineering, technology, business. A public body or a small firm hiring by word of mouth is not in this number and neither are they in anybody else's.

A prediction. We have one measurement, taken once. We will publish it again.

Why we are giving it away

Because the alternative was to put it behind a form.

Our own product measures the same thing: salaryMin on a posting means the employer published this, and it is deliberately kept separate from our salary estimate, which is a median of published bands in the same country. Merging the two would make the transparency figure unmeasurable and every downstream reader an unwitting liar — including us. That separation is why we can count this at all.

The full set of platform figures, each with the query it came from and the date it was last run, is on our facts page. If you want to argue with a number, you should be able to see how it was made.