The Global Talent route does not care where you work. It cares what you can prove. There is no sponsor, no salary threshold, no job offer and no English test at the visa stage, and in exchange the whole weight of the application lands on a folder of documents you assemble yourself. Ten pieces of evidence, three letters, a thousand words and a CV. That is the entire case. This is a checklist for building that folder, slot by slot, written for engineers, founders, researchers and product people rather than for lawyers.
How the assessment actually works
Worth ten minutes before you open a document. Almost every avoidable refusal comes from somebody who assembled impressive material and never mapped it to the structure the assessor is working through.
One mandatory criterion and two of four
The mandatory criterion is recognition. For Exceptional Talent you show you have been recognised as a leading talent in digital technology in the last five years. For Exceptional Promise you show you have been recognised as having the potential to become one over the same period.
Then you pick two of four optional criteria. Innovation, meaning a proven track record as a founder or senior executive of a product-led digital technology company, or as an employee working on a new digital field or concept. Recognition for work beyond your own occupation that has advanced the field. Significant technical, commercial or entrepreneurial contributions to the field. Or exceptional ability demonstrated through academic contributions published or endorsed by an expert.
You choose the two. Nobody chooses them for you, and choosing badly is the difference between a pack that assembles itself and one you spend three months forcing.
Technical or business, and why the label matters
The application asks you to identify as a technical applicant or a business applicant. Technical covers people who build — engineers, architects, data scientists, security specialists, researchers. Business covers commercial, product, operations and investment roles inside technology companies.
The criteria are the same but the evidence that satisfies them is not. A technical applicant is expected to show what they built and what it did. A business applicant is expected to show revenue, growth, funding and market outcomes attributable to their decisions.
Hybrid careers are common and awkward. If you have moved from engineering into product leadership, pick the label that matches the last three years and where your strongest evidence sits, then keep the whole file consistent with it.
Talent or promise, and the three-year difference
You declare which category you want to be assessed under. Exceptional Talent suits people with more than five years in the sector and a record of impact that others outside their own company can see. Exceptional Promise suits people earlier in their career who can show trajectory.
The consequence is settlement. Exceptional Talent leads to indefinite leave to remain after three years. Exceptional Promise takes five. That gap is worth two years of your life and several thousand pounds in health surcharge, so it deserves a serious conversation rather than a reflex.
Being ambitious about the category and failing is expensive. Being cautious and succeeding puts you on the longer road. The honest test is whether your recognition evidence would convince somebody who has never heard of your employer.
If the two-year difference is what is driving your decision, it helps to see the dates rather than the theory, and you can calculate your ILR eligibility date under each category before you commit to one.
The ten slots, and how to spend them
Ten documents is the whole budget for the evidence section. The personal statement, the CV and the three letters sit outside it and do not use up slots. Every one of the ten counts, including the weak one you included because you had space.
Two for the mandatory criterion, at least
The mandatory criterion asks for recognition, so the evidence has to come from outside you and outside your line manager.
Things that work here. A conference talk at a recognised event, evidenced by the programme and an invitation rather than by your own slides. Press coverage in a publication that has editorial standards. An industry award with a named judging process. A significant open source project with visible adoption. Being asked to advise, judge, mentor or sit on a technical committee. A patent granted or filed with your name on it.
Things that read as thin. A LinkedIn post with engagement. A company blog you wrote. A meetup you organised for twelve people. A certification. A university degree on its own, at any level.
Two is the floor and three is often better, because recognition is the criterion where assessors look hardest for independence.
Four for your two optional criteria
Two pieces per optional criterion is the minimum, and they should be different in kind rather than two versions of the same thing.
For innovation, that might be a product you led and the commercial outcome it produced. For contribution, the technical work and the measurable result of it. For academic contribution, the paper and the evidence of its reception rather than the paper twice.
Label each document clearly with which criterion it addresses. Assessors are reading a lot of files and will not go hunting for the connection you left implicit. A one-line summary at the top of each piece explaining what it is and what it proves costs you nothing and changes how the whole pack reads.
What to do with the remaining four
You do not have to use all ten and there is a real argument for not doing so. Recent guidance has leaned towards fewer, stronger documents, and reviewers openly say that dense or hard-to-navigate applications go against the applicant.
Where you do use the spare capacity, use it to deepen rather than to widen. A third piece supporting your strongest optional criterion is worth more than a first piece supporting a third criterion you were never going to rely on.
The one thing not to do is pad. A weak document does not sit neutrally in the file. It tells the assessor something about your judgement, and about what you thought was worth including.
What a strong piece of evidence looks like
This section is the actual checklist. Run every document you are considering through it and discard the ones that fail.
The format rules that get files rejected on sight
Each piece of evidence is a maximum of three A4 sides. You may not combine several items into one document to fit more in. Everything must be typed. Everything is uploaded through the application platform.
Evidence has to come from the last five years. Anything older is out of scope, and anything obviously produced close to your submission date invites the question of whether it exists only because you were applying.
Screenshots need restraint. One or two per page, cropped, legible, with a caption explaining what is being shown and when. A page of unlabelled dashboard images proves nothing to somebody who has no context for your product.
Naming your own contribution rather than your team’s
This is where technically strong applicants lose. A document describing a system that handles two hundred million requests a day tells the assessor about the system. It says nothing about you unless it says what you did.
Write the contribution explicitly. You designed the sharding strategy. You led the migration. You wrote the component that reduced latency by a specific amount. You made the architectural call that let the team ship. Then have somebody senior confirm it in writing, because a claim about your own role is worth much more when a third party repeats it.
Where you were one of several people, say so and describe your part honestly. Assessors have seen enough applications to recognise an inflated claim, and one of those undermines the credible parts of the same file.
Numbers, sources and dates
Every document should answer four questions without being asked. What is this. When did it happen. Who says so. What changed as a result.
Numbers carry the most weight, and current guidance leans hard towards commercial and financial proof rather than reach alone. Users, downloads and stars are worth something. Revenue, contracts, cost saved and funding raised are worth more.
If you are relying on a business you founded, projections and a business plan will not do it. Audited accounts, or formal accounts covering more than twelve months, are what is expected. Articles of association on their own prove that a company exists rather than that it worked.
What does not count
A job title at a well-known company. A high salary. A degree, a masters or a bootcamp certificate. Years of experience by themselves. An offer letter from a UK employer. Anything under an unbroken non-disclosure agreement that you cannot describe in any form.
Also excluded in practice, though people keep trying. Duplicated evidence used twice for two criteria. Documents in another language without a translation. Anything you wrote about yourself and presented as external recognition.
A useful test before you include something. If a stranger read only this document, would they learn something specific about your work that they could verify. If the answer is no, it is not evidence.
The three letters
Recommendation letters have come back into focus after a period of being treated as a formality, and they are now one of the clearest signals in the pack. They are also the part with the longest lead time, so start here rather than finishing here.
Who should write them
Three letters from three different, well-established organisations, written by three different people. Each author should be a senior figure in digital technology with detailed personal knowledge of your work.
Seniority alone is not the point. A chief technology officer who barely knows you writes a worse letter than a principal engineer who reviewed your code for two years. Assessors can tell the difference immediately, because the specific letter contains things nobody could invent.
A UK-based author among the three helps, because it demonstrates that your reputation reaches the market you are asking to enter. It is not a requirement and a weak UK letter is worse than a strong overseas one.
What has to be in each one
Each letter runs to a maximum of three pages. It must be written specifically for this application rather than repurposed from a job reference, and it should say so.
It needs the author’s name, role, organisation, telephone number and email address, on headed paper with the registered address where the organisation has one. It must be signed and dated. It should come with the author’s own CV or other proof of their standing in the field.
The content has to do three things. Explain how the author knows you and for how long. Describe your specific achievements with detail only somebody close to the work would have. State plainly why they regard you as a leader, or a potential leader, in digital technology.
Letters that read as templates, or that closely resemble evidence submitted by other applicants, may be treated as unacceptable or worse. This matters if you use a service that supplies letter drafts to many clients.
How to brief a recommender without writing it for them
Send them a short pack rather than a draft. Your CV. A one-page summary of the two or three achievements you want them to be able to speak to, with the numbers attached. The criterion you are relying on their letter to support. The formatting requirements. And a deadline three weeks earlier than your real one.
Then let them write it in their own voice. A letter that sounds like the person who signed it is the point. If three letters arrive with the same sentence structure and the same adjectives, you have created a problem rather than solved one.
Read each one when it arrives and check it against the requirements before you thank them. Going back a second time is awkward. Submitting a letter with no phone number on it is worse.
The personal statement and the CV
Neither of these counts towards your ten slots, and both are routinely treated as afterthoughts by people who have spent two months on the evidence. That is backwards. These two documents are what the assessor reads first, and they set the frame everything else is read through.
A thousand words with a job to do
The personal statement runs to a maximum of a thousand words and is meant to explain the contribution you will make to the UK digital technology sector.
A structure that works. Who you are and what you build, in two sentences. What you have achieved, mapped explicitly to the criteria you are relying on, with the evidence numbers referenced. What you intend to do in the UK, concretely. Why the UK rather than anywhere else, and why now.
Be specific about the plan. Naming the companies you have spoken to, the sector you will work in, the problem you intend to work on and the city you will base yourself in reads as somebody with a purpose. Saying that you are excited about the UK technology ecosystem reads as somebody filling space.
Three pages of CV, weighted to the last five years
The CV is three A4 sides at most, typed, with clear dates and no gaps you cannot explain.
Weight it towards the last five years, because that is the period being assessed. A decade of earlier roles compressed into four lines is fine. The last three positions need enough detail that the achievements in your evidence have somewhere to attach.
Include publications, patents, talks and open source work as a section rather than scattering them through the roles. And make sure every date, title and company name matches what appears in your letters and your evidence, because inconsistencies between documents are noticed and they cost credibility.
Expert note from the LawSentis immigration team
The pack we see most often is technically outstanding and evidentially quiet. Somebody has spent nine years building things that matter and has almost nothing outside their own employer to show for it, because nobody in that career ever needed external proof before. That is a fixable problem, but not in three weeks. Give a talk. Publish the thing you built internally, in a stripped-down form your employer will approve. Get the patent filed. Say yes to the podcast, the panel and the mentoring programme. Six months of deliberate visibility turns a borderline Promise case into a comfortable Talent one, and the cost of doing it is far lower than the cost of a refusal followed by a rebuild. The applicants who struggle are rarely the ones without achievements. They are the ones whose achievements have never left the building.
Evidence by profile
The criteria are written once but they land very differently depending on what kind of technologist you are. Four common profiles and where each one usually finds its evidence.
Engineers and architects at established companies
Your strength is depth and your weakness is attribution, because everything you built has a company logo on it rather than your name.
Build the pack around specific systems and specific decisions, evidenced by internal documents your employer will release, performance data with the numbers intact, and letters from people who watched you make the calls. Patents help enormously here because they are external, dated and carry your name.
Then push at least two pieces outside the company. A conference talk, a published technical article, an open source contribution, a standards body role. Without those, the recognition criterion is doing all its work inside one organisation, and that is exactly where it is weakest.
Founders and early employees
Your strength is outcomes and your weakness is proving them. The route treats founders well but expects financial substance rather than narrative.
Accounts covering more than twelve months. Investment agreements or term sheets. Customer contracts. Revenue figures with something behind them. User numbers where they translate into commercial reality. Coverage from publications that write about companies rather than press releases.
If the company failed, that is not disqualifying and should not be hidden. Businesses dissolved in the last five years where you were a founder or senior executive are expected to be disclosed. A candid account of what you built, what happened and what you learned reads far better than a gap somebody notices later.
Open source, research and community work
This profile is often stronger than the person thinks, because almost all of the evidence is already external and dated.
Repository metrics with real adoption. Citation counts. Papers with the venue and its standing explained. Maintainer status on a project other companies depend on. Invitations to review, speak or sit on programme committees. Grants and fellowships.
The thing to add is commercial translation. Explain who uses the work and what it enables, because the assessment is about impact on the digital technology sector rather than about academic merit for its own sake.
People at consultancies and agencies
The hardest profile, and worth being honest about. Several of the criteria are framed around product-led digital technology companies. Pure consultancy, outsourcing, services and businesses that use technology without building their own product often sit outside that framing.
If this is you, look for the parts of your work that were product-shaped. Internal tools that became products. Platforms you built that the client now sells. Reusable technology you created rather than projects you delivered. Then lean harder on the recognition and academic criteria, which are not tied to the company type in the same way.
This is also where an outside read is worth the most, because the question of whether your history fits the criteria at all is one you should settle before you spend three months building, and our Global Talent visa support starts with exactly that assessment rather than with document drafting.
Timeline, cost and what happens if it goes wrong
The last section, and the one that decides whether any of the rest of it happens. Most people who never apply are people who underestimated the lead time on the parts they did not control.
A build schedule that works
Three to six months is normal for a serious pack, and the shape of it matters more than the total.
Months one and two go on the letters, because they depend on other people. Approach the authors, send the brief, chase gently, and treat anything that arrives in the first fortnight as a pleasant surprise. In parallel, list every candidate document you could use and mark each one against a criterion.
Month three is selection and drafting. Cut to ten, write the summaries, draft the personal statement, rebuild the CV. Month four is checking, which sounds excessive until you have compared three letters, ten documents and a CV for consistency and found four discrepancies.
The Home Office runs a single first-stage application and passes it to the endorsing body, so nothing goes to Tech Nation directly. Endorsement decisions typically take four to eight weeks. Once endorsed you have three months to make the visa application, which is then decided in around three weeks from outside the UK or eight from inside.
What it costs and how to sequence it
The Home Office charges £561 at the endorsement stage and £205 at the visa stage, £766 in total for the main applicant. Anyone qualifying through a listed prestigious prize skips endorsement and pays the £766 in one go at the visa stage.
Then the health surcharge, at £1,035 per adult per year and £776 per child, paid upfront for the whole period you request. You choose the visa length in whole years between one and five, which is a genuine lever. A three-year grant costs £3,105 in surcharge and lines up neatly with the Exceptional Talent settlement point. Five years costs £5,175 and saves you an extension.
The sequencing point that matters. If the endorsement is refused, the endorsement fee is gone but you have not yet paid the visa fee or the surcharge, so the financial exposure at the first stage is £561 plus whatever you spent on preparation.
Refusal, review and the second attempt
Endorsement refusals come with feedback, and there is a review procedure. Understand what it is before you rely on it. A review generally examines whether the original decision was made correctly under the criteria and the process. It is not an opportunity to rebuild the case with the evidence you wish you had submitted.
So for most people the realistic response to a refusal is a fresh application with a materially different pack. Read the feedback against your file line by line and work out which criterion failed and why, because reapplying with the same evidence in a different order produces the same answer.
There is no cooling-off period and no cap on the number of endorsements the route can issue. What there is, is time and money, and both are better spent on six months of building evidence than on a second submission of the same material.
FAQ:
- Question — Can I apply from inside the UK if I am already here on another visa?
Answer — In most cases yes. Global Talent permits switching from within the UK from a range of routes, including Skilled Worker, Student, Graduate and Innovator Founder, and there is no requirement to leave and apply from abroad. Visitors cannot switch. The practical advantages of applying in-country are that your existing permission continues while the second-stage application is pending, and that you may already have UK-linked evidence and a UK recommender, both of which strengthen the pack. The practical disadvantage is timing, since in-country decisions typically take around eight weeks against three from overseas, so start early if your current permission is running down.
- Question — Do I need a UK job offer, a UK employer or any UK connection before applying?
Answer — None of those is required and the absence of them is the point of the route. You do not need a sponsor, an offer, a minimum salary or an employer of any kind, and once granted you can work employed, self-employed, freelance, or run your own company, in any combination. What does help is showing that you have thought concretely about the UK rather than treating it as a destination. Naming the sector, the problem and the people you have spoken to in your personal statement is not an eligibility requirement, but it reads very differently from a general statement of enthusiasm, and assessors notice the difference.
- Question — My employer will not let me disclose revenue or user figures. What can I use instead?
Answer — This comes up constantly and it is workable. Ask for a letter from a senior person at the company confirming the figures and the impact in general terms, on headed paper, rather than releasing the underlying data. Percentages and orders of magnitude are usually acceptable where absolute numbers are not, so a confirmed forty per cent reduction in processing cost does the job without exposing the cost base. Internal recognition documents, promotion records tied to specific work, and performance data with the commercially sensitive columns removed all work. What does not work is asserting the impact yourself with nothing behind it, so if the company will confirm nothing at all, build the criterion somewhere else instead.
- Question — If I am endorsed under Exceptional Promise, can I move to Exceptional Talent later?
Answer — Not by conversion. The category is fixed by the endorsement you hold, and there is no upgrade mechanism attached to it. What you can do is make a fresh endorsement application under Exceptional Talent once your record supports it, and if that succeeds you would be on the three-year settlement timeline attached to the new endorsement. Whether that is worth doing depends on how far into the five years you already are and what a second application would cost you in fees and preparation. For many people two or three years into a Promise endorsement, the answer is to stay put. For somebody in their first year whose career has moved quickly, it can be worth the second run.
- Question — Can my partner and children come, and will we all settle at the same time?
Answer — They can come, and no, you will probably not settle together. Partners and children under eighteen apply as dependants using your endorsement, are not assessed against any of the criteria themselves, and can work and study once here. Each pays the same £766 application fee and the health surcharge for the full period. The timing catch is that a dependent partner normally needs five years of residence in that capacity before qualifying for settlement, even where you qualify after three. That means the main applicant can reach indefinite leave to remain two years ahead of their partner, who will usually need a further grant of permission to close the gap. Plan the visa lengths for the whole family together rather than optimising your own.