Job Market Signal — June 2026

June 7 was the day every EU member state was supposed to have written the Pay Transparency Directive into national law. June is the first full month with that deadline behind the market. This edition reads what happened next — which, in the postings, is almost nothing.

One housekeeping note first, because it matters for every number below. Seven new sources joined the crawl during June — boards in Portugal, Poland, Hungary, Spain, Austria, the Baltics and the Netherlands. A naive June-versus-May comparison would mostly measure that change, not the market. So every month-over-month figure in this edition is computed on the constant set of 43 sources present in both months. When we say something moved, it moved.

The deadline came and went. The needle didn’t.

On identical sources, EU-weighted salary disclosure went from 30.2% to 30.0%. Flat.

Last month’s edition closed with the question worth watching: would the passed deadline pull European employers toward transparency? June’s answer is no — or at least, not yet. On the constant source set, the EU-weighted disclosure rate sat still. The United States, under no EU mandate at all, disclosed pay in 60.3% of postings — twice the European rate.

Inside the flat average, two countries actually moved — in opposite directions. Germany fell from 17.3% to 15.4% on identical sources, in the month its (still unwritten) transposition law became overdue. The Netherlands rose from 21.6% to 26.5% — the largest genuine gain of any major market, and it comes with the Dutch transposition pushed to 2027. If Dutch employers are moving ahead of their own law while German ones drift the other way, that says something the directive’s text doesn’t: disclosure follows market norms first and mandates second.

Remote is shrinking — measurably, this time

Remote fell from 5.4% to 4.3% of postings on identical sources. Roughly a fifth of the remote market, gone in a month.

Last month’s edition could only say that remote is rare and concentrated. This month, with a constant footprint, it can say something stronger: remote is declining. 93.8% of June’s postings read onsite. A seeker filtering for remote work is now searching inside one posting in twenty-three.

The concentration rule still holds — and it matters more when the pool shrinks:

The ordering reshuffled since May — support and customer service now leads, marketing dropped from first to fifth — but the practical rule survived intact: the role family you target moves your remote odds far more than any company’s policy does.

Less coordination, more analysis

Project-management skills fell about a fifth in relative terms. Analysis skills rose. Each cluster moved as a whole.

Industry shares barely moved in June — no sector shifted more than a percentage point. The movement is inside the postings, in what they ask for. Measured as a share of all skill mentions on identical sources, every project-management label in the ESCO taxonomy fell together (operations management −27%, managing project metrics −23%, project management −19%), and every analysis label rose together (cost-benefit analysis +31%, financial analysis +20%, customer-needs analysis +17%).

One month of movement earns a watchlist entry, not a conclusion. But whole clusters of related skills moving in lockstep is a stronger signal than any single label — and one plausible reading is that organisations are hiring to understand their situation before hiring to run more of it. If July repeats the pattern, this becomes the story of the summer.

The entry-level wall has a second brick

Entry-level roles held at exactly 2.5% of the market. And they disclose pay least of any seniority.

A second full month at exactly one posting in forty settles it: the entry-level wall is structure, not noise. June adds a detail that makes it worse. Entry postings disclose salary in 26.8% of cases against 36.8% for mid-level roles — the people with the least information to negotiate with are given the least of it.

And the AI question, quantified for the first time: in June, 0.13% of all postings name a machine-learning or artificial-intelligence skill from the ESCO taxonomy — roughly one posting in 770. Even inside data and AI roles, only 0.4% do. In May the figure was 0.16%. Whatever is happening to work because of AI, employers are not yet writing it into what they hire for — and the narrow entry door predates the model wave. This number now recurs every edition, so when it finally moves, you’ll see it here.

Signatures move. That’s why we read them monthly.

Norway's care signature held. The US deepened around behavioural health. France turned logistics-heavy.

Some fingerprints are stable: Germany’s mechatronics and Norway’s care-climate-pedagogy cluster held from May. Others moved. The US signature deepened around behavioural health — four of its five most over-indexed skills sit in clinical and therapeutic work. France’s turned logistics-heavy, and Switzerland’s biomedical lead from May gave way to hospitality and ICT systems work. For a job seeker, the reading stays the same: the market that rewards your specialty is not always the one next door — and it isn’t always the same one it was last month.

What to watch into July

Three questions carry over. Whether Europe’s disclosure rate finally reacts as transposition laws start landing — Germany is now the country to watch, moving the wrong way. Whether the analysis-over-coordination rotation repeats, turning a one-month move into a demand shift. And whether remote’s decline continues — a second month down would make it a trend, not a wobble.

One promise kept, honestly: last month we said trend lines would wait for a frozen source footprint. The footprint grew instead, so every trend here uses the constant-source method described above — and one planned feature stays unpublished, because May’s day-of-week posting pattern did not replicate in June. It returns when it’s stable, not before.

Methodology: every figure comes from Kitsuno’s production crawl, classified by the Extractor agent using the ESCO skills taxonomy and an internal role-family scheme. The dataset covers 420,626 unique postings collected across June 2026 from 50 sources in 109 countries; 83.0% carry an identified location. Seven sources joined the crawl during June, so all month-over-month comparisons are computed on the constant set of 43 sources present in both May and June. The AI-skill measure counts postings naming at least one ESCO skill whose title contains “machine learning” or “artificial intelligence” — a deliberately strict definition of named demand. Salary disclosure is pattern-detected and extraction varies by market: France reads high, the UK and Netherlands likely low, so movements are more reliable than absolute levels.

For how the pipeline works, read How we measure job-market signal. For the live numbers, visit Market Pulse.