Issue 016: One in eight never arrive

On Tuesday 4 August Qualifications Scotland published its results and UCAS released the first acceptance figures of the 2026 cycle (source). A level results follow on Thursday 13 August, and for the ten days after that, acceptance numbers will be the sector's favourite statistic. The Spotlight below counts the students who accept a UCAS place and never enrol. At 35 UK providers it is one acceptance in eight, in every cycle since 2021.

The enforcement news moved too. On 5 August the Home Office revoked Bloomsbury Institute's student sponsor licence, the first revocation since the compliance thresholds tightened on 1 June 2026 (source). And judgment in the High Court challenge to the visa brake is expected by mid August. Details on both are below.

HEPI published Policy Note 73 on 6 August. Vicki Stott, who used to run the Quality Assurance Agency, wants a formal accountability framework for franchised and partnership provision (source). Her numbers come from Tom Richmond's April 2026 HEPI paper A degree of regulation (source). Students in franchised higher education more than doubled in three years, to 108,600 by 2021/22, and reached 135,850 in 2022/23. Most were taught by providers not registered with the OfS. And 53% of the Student Loans Company fraud detected in 2022/23 sat in franchised provision. Stott's sharpest idea is to treat large scale franchising, in her words, as a licensed activity rather than a routine commercial choice, with the OfS running an approval regime for the riskiest arrangements.

The stories this week point the same way: attention is shifting from whole institutions to the routes students arrive through. Bloomsbury lost its licence because of what happened to the students it accepted. Stott wants the franchise route licensed, and that is the route carrying much of the growth in Issue 15's invisible intake. On 8 August Jim Dickinson showed the loans route, where agents can earn £2,000 a student, and a £15,000 BMW at 200 enrolments, for recruitment rather than completion (source). And the Spotlight below shows the sector's favourite number, the UCAS acceptance, overstating arrivals at 44 providers. Each found the same thing: the numbers behind recruitment are weaker than the decisions built on them.

In the 2024 cycle, 44 UK providers accepted more international students through UCAS than they enrolled from every route combined: 32,165 UCAS acceptances against 27,360 HESA entrants, a gap of 4,805 students (source). These are not students lost to rival routes; the HESA side counts every route into the provider, agents and direct applications included. The acceptances simply exceed the arrivals.

This is not a 2024 story. The same count was 46 providers in 2021, 50 in 2022, 51 in 2023 and 44 in 2024, with the gap at 4,770, then 4,735, then 4,475, then 4,805. Across the four cycles, close to 19,000 international acceptances never became students at the accepting provider. Thirty five providers have been over the line in at least three of the four cycles, and at that group arrivals from every route combined came to 87.1%, then 88.6%, then 88.4%, then 86.9% of UCAS acceptances. One acceptance in eight is not matched by any arrival, and the rate barely moves.

Twelve of the 35 are Russell Group members. The University of Edinburgh accepted 2,670 international students through UCAS in the 2024 cycle and recorded 1,975 non-UK first degree entrants in HESA's 2024/25 data, a gap of 695 and the largest in the group. UCL accepted 4,590 against 4,105 entrants. Cardiff accepted 1,225 against 870, Glasgow 1,385 against 1,050. Every one of those four has shown the same gap in every cycle since 2021. Kingston, Bath and Exeter complete the persistent names with the largest 2024 gaps.

The method is the one built and verified for Issue 15: UCAS end of cycle provider acceptances for international applicants (source) matched against HESA first degree non-UK entrants (source) for 170 providers across the 2021 to 2024 cycles. Issue 15 used the match to show most non-Russell Group intake arriving outside UCAS. This issue reads the same file in the other direction: the providers where the UCAS number overstates the intake.

Deferrals and January intakes do not explain it. A student who accepted in 2024 and started in autumn 2025 leaves the 2024 count, but 2023's deferrers join it, and at a steady deferral rate the two broadly cancel. A provider showing the gap once might be timing; a provider showing it four cycles running is not. The sums stay small even on hostile assumptions: at a deferral rate of 10%, the largest year to year swing in any named provider's acceptances could move its gap by about 60 students, against gaps in the hundreds. January starters count on the enrolment side, which makes these gaps smaller, not larger. What remains is students who accepted a place and never appeared in that provider's intake: visa refusals after acceptance, students who switched provider late, students who never travelled, and withdrawals inside the first two weeks, which HESA's counting rules exclude. None of those appear in any published series.

Two things this comparison is not claiming. First, no admissions system converts every acceptance; some loss is normal everywhere. Second, because it sets one dataset against another, UCAS acceptances against HESA arrivals, it can only see no-shows at providers that recruit nearly all of their international undergraduates through UCAS. At a provider with a large intake outside UCAS, the same no-shows are hidden inside the bigger arrival number. So 4,805 is what the two datasets can show, not the full size of the loss. Across the whole sector, the number of acceptances that never become students will be higher. One removal for the same reason: Greenwich showed the largest single-year ratio in 2024, but its swings reverse from year to year, which looks like intake timing rather than no-shows, so it is not cited here.

International students paid £12.7bn in tuition fees in 2023/24, 47% of all UK tuition fee income, on HESA data quoted to Parliament in March 2026 (source). Across 732,285 non-UK students that year (source), that is about £17,300 a head. It is a deliberately conservative average, blended across every level of study, and first degree international fees at the named providers mostly run higher. At £17,300, the 4,805 no-shows of the 2024 cycle are about £83m in first year fees, and roughly £250m across a three year degree. Every cycle, at the most conservative price. Clearing refills some of those places, so this is the size of the leak, not a claim that every pound stays lost. A leak this stable is a budget line, and almost nobody carries it as one.

The constant also makes a forward claim possible. When UCAS publishes the first full 2026 acceptance figures on Thursday 13 August, apply the rate: at these 35 providers, about one acceptance in eight will not become a student. We will mark that claim against the enrolment data when it arrives, as Issue 12 did with its study visa floor.

For a Head of International, the operational question is not whether the sector's constant is exactly 13%. It is whether your own conversion beats it, and whether you can name the parts: how many of your accepted students were refused visas, how many switched away after acceptance, how many simply never came. Providers that can separate those three can fix them separately. Providers that only count acceptances will keep budgeting for students who were never coming.

Things to think about this week

Acceptances are the only recruitment number available in real time, which is why they will dominate every Vice-Chancellor's update between Thursday 13 August and the start of term. Arrivals are the number that pays the bills, and it is published more than a year later by a different agency on a different definition. The no-show constant lives in the space between the two.

One in eight is a sector average of a stable, measurable, unmanaged leak. The providers in the chart above are not weak recruiters; several are the strongest brands in the country. They are institutions where nobody's job depends on the difference between an acceptance and an enrolment. This week, while the acceptance numbers roll in, is the right moment to find out who owns that difference at yours.

If this was forwarded to you, subscribe to receive every Tuesday at 7am UK time.