Sponsor register data & methodology

The register

The list of licensed sponsors comes from the UK Home Office's Register of Licensed Sponsors (Workers), published on GOV.UK as a CSV under the Open Government Licence v3. It names every organisation currently permitted to sponsor workers on routes such as Skilled Worker, Global Business Mobility and the Temporary Worker routes, along with the town on the licence, the licence type and rating, and the routes it covers.

The Home Office republishes the register frequently — in practice on most working days. This site re-reads GOV.UK daily, compares the new edition against the last one, and records every addition and removal with the date it was observed. That dated history is the part GOV.UK does not keep: the official file only ever shows the current state, so once an organisation drops off, the register itself no longer shows that it was ever there.

Register fields are republished as-is. Organisation names are shown in title case for readability, and the underlying value is unchanged.

Active, revoked and ceased

A sponsor is Active when it appears in the current edition of the register. When an organisation disappears, its page stays online and the status becomes one of two things.

Revoked means the organisation has left the register while the company itself is still a live entity at Companies House. Ceased means it has left the register and Companies House shows the company as dissolved, in liquidation, in administration or otherwise ended.

This distinction is drawn from Companies House status alone, and only ever on positive evidence that the company ended — anything unrecognised stays classified as Revoked. That matters because the two look identical in the register: GOV.UK publishes no reason when an organisation leaves, so a licence withdrawn after enforcement action and one voluntarily surrendered by an employer that no longer needs it are indistinguishable in the data. This site reports that a licence ended, and when. It does not claim to know why.

Companies House enrichment

Where a sponsor can be matched to a Companies House record, the site tracks changes filed since the licence appeared: new people with significant control, transfers of shares, directors joining and leaving, registered-office moves, name changes, and above all whether the company is still trading.

This exists because ownership changes carry Home Office consequences. A takeover or a sale of controlling shares can require an entirely new sponsor licence; most other reportable changes must be reported through the Sponsorship Management System within 20 working days of the event. Each change shown on a sponsor's page is mapped to the duty it triggers and the deadline that follows from it.

Each sponsor is re-checked roughly every three months, and the date of the last check is shown alongside the result.

The revocation score

The revocation score answers one narrow question: how likely is it that this licence is revoked while the company is still trading, within roughly the next seven months. It is a statistical model, not a Home Office figure, and it is fitted on real outcomes — 4,574 sponsors that actually left the register.

The base rate is about 3.7 sponsors in 100. The model converts a sponsor's Companies House facts into a calibrated probability, and the 0–100 score is that probability relabelled so each band owns a fixed slice of the scale. The band is therefore always recoverable from the score, and nothing is added on top.

There are two versions. The full model uses 74 signals from the Companies House record and scores AUC 0.741 on held-out data, where 0.5 would be a coin flip; working the riskiest tenth of the register catches 17.4% of all revocations, about 5.5 times the base rate. A reduced model using only the six signals available from the register row is used until a Companies House check upgrades the row; it is substantially weaker at AUC 0.584 and by construction can never enter the top band.

What predicts revocation turns out to be mostly what a sponsor is rather than what it does: sector and company age dominate, and the only corporate action to survive a strict screen was moving the registered office.

What the score deliberately excludes

Two signals that passed the statistical screen were removed on inspection, because they were measuring the outcome rather than predicting it.

• The UK Expansion Worker route. That licence type is provisional by design and simply expires, which the study's outcome definition had counted as a revocation. Its apparent 5.8x risk was scheduled expiry, not enforcement.
• Live job adverts. A removed sponsor cannot advertise, so the outcome determined the feature and it scored as strongly protective — pure leakage. Job-advert counts are severed from the scoring path entirely.

How much to trust it

The study behind the model was built not to fool itself: every fact was reconstructed as of the scoring date to avoid look-ahead, the replay was diffed against the production checker, 564 candidate signals were screened on data splits that never met, the mathematics was unit-tested against textbook values, and every published figure was generated by script rather than transcribed.

The honest limits are these. GOV.UK publishes no removal reason, so surrender and enforcement look the same in the training data. And licences are revoked for illegal working, underpayment and record-keeping failures — none of which appears anywhere at Companies House. AUC 0.741 is the ceiling for this data source; improving on it needs different data, not better mathematics.

A high score is a reason to look closer at a sponsor, not evidence of wrongdoing. Roughly 80 in 100 sponsors in the highest band do not lose their licence in the window the model covers.

Job listings

Live vacancies are harvested daily from Reed, Adzuna, Jooble and Careerjet and matched back to register entries by employer name. Matching is by name, so it is approximate: a listing shown against a sponsor means an advert from an employer of that name, and a vacancy appearing on a sponsor's page does not by itself mean the role is offered with visa sponsorship.

Listings are never deleted when they close — they are marked expired and kept, so a sponsor's hiring history stays visible.

What this site is not

This is not the official register, and it is not legal advice. It is a searchable copy maintained by Borderless Immigration, an OISC-regulated UK immigration provider, with dated change history and Companies House context added.

For anything decision-critical — an immigration application, an employment decision, a compliance response — confirm the current position on GOV.UK and take advice.