July 2026

Executive summary

Three things happened to the figures this month before anything that actually happened in the labour market was reported. ONS revised Labour Force Survey data back to January to March 2020 after a seasonal adjustment review. Its two main measures of employee numbers moved in opposite directions over the same quarter, the survey up 115,000 and the tax records down 30,000. And a collection error in May required its own impact assessment. This is the second briefing running in which a figure we published has since moved. Revisions of this kind mostly affect quarterly readings, so this briefing leans on annual comparisons.

On the quarter, unemployment fell 0.1 percentage points, employment rose 0.1 and inactivity fell 0.1. All three point mildly the right way, all three sit inside the range where a revision could reverse them, and we draw no firm conclusions from them. The useful reading is in the composition.

Youth unemployment rose again, from 15.9% to 16.4%, and is now more than two points above where it was a year ago. Vacancies at businesses with one to nine employees are down 9.3% over the year, while employers with 2,500 or more are up 3.2%. Hospitality and retail have lost more than 140,000 employees. Where vacancies are growing, they are growing in education and construction, sectors that generally require a qualification or a licence to enter. The first rung is not only crowded, it is being removed. Real regular pay grew 0.3%, and private sector real wages have fallen since October. Alan Milburn's recommendations on young people and work have slipped again, to early autumn, close to a year after the review was commissioned.

The most useful finding sits underneath the headline. Recruitment survey data shows permanent placements close to stabilising, while total demand for staff fell at its quickest rate since January. Both can be true. Candidate availability is still rising, largely because of redundancies, while vacancies keep falling, which makes each remaining vacancy easier to fill. What has improved is matching, not hiring. It also matters who that reaches. Matching recently redundant people into fewer roles helps those who were in work weeks ago, and can worsen the position of someone out of work for a year by putting stronger applicants in the same queue. That gives the divide we described in June a mechanism rather than only a pattern. Two measures would have to move before we would be suggesting a positive turn: total demand for staff, and permanent vacancies. Both are still falling.

Career Allies and the Stay Nimble platform continue to scale as set out previously, and this month's evidence sharpens that argument rather than changing it. Coaching outcomes this period took months, and in some cases years, to arrive. We also correct two observations from earlier issues.