Ask any employer what they want from a graduate, and the list is strikingly consistent: not just knowledge, but judgement, communication, collaboration and resilience.
Ask the same employer how confident they are that a CV, a personal statement or a set of exam grades actually proves a candidate has those things, and the answer changes.
Not very.
TestGorilla’s 2025 State of Skills-Based Hiring report found that 53% of employers say working out whether a candidate has the right skills, soft or technical, is the hardest part of hiring. Roughly nine in 10 UK and US employers report the same underlying problem with a CV: it cannot tell them whether what is written on it is actually true.
Here is the more uncomfortable number. Harvard Business School and the Burning Glass Institute tracked more than 11,000 job postings between 2014 and 2023, at firms that had publicly dropped degree requirements to open roles up to skills-based hiring. Fewer than one in 700 new hires were people who would previously have been screened out for lacking a degree.
The job adverts changed. The hiring barely did. That gap, between the policy an organisation announces and the decisions it actually makes, is not a hiring footnote. It is the central problem in how we recognise capability at all: strip out the credential, and most employers still have no reliable way to check what should replace it, so they quietly keep leaning on the signal they know how to read.
Schools are now facing a strikingly similar gap.
I wrote here last month about why schools are looking beyond exam results for a fuller picture of what a student can do. What I did not address is the harder question sitting underneath that ambition, and it is one I keep returning to in conversations with school and university leaders across the UK, the Middle East and Asia.
If a school says a student can collaborate, communicate or think critically, how do we know that is true? A reference letter is one person’s judgement. A personal statement is self-reported, dressed in a teacher’s approval. A skills profile is often a form filled in once a year, more effort than evidence. Each of those documents is a claim. None of them is evidence.
None of that mattered much when nobody was relying on those claims to make a hard decision. They are relying on them now, at exactly the moment those claims became easiest to produce without any of it being true. In an era of generative AI, a reflective essay or a glowing personal statement can be manufactured in seconds – with zero underlying evidence that any of it actually happened. Universities and employers are both quietly grappling with the same problem, from opposite ends of the same pipeline.
None of this is a reflection on teachers, who have always assessed judgement and character as well as anyone could reasonably ask. It is a reflection on the systems around them, which were mostly never designed to produce evidence that could survive being questioned by someone who was not there.
So what does verification actually look like, done properly?
In an era of generative AI, a reflective essay or a glowing personal statement can be manufactured in seconds – with zero underlying evidence that any of it actually happened
Not more exams. That would only reward a different kind of performance, better trained for, not better proven.
It means designing assessment so the evidence of a skill and the evidence of who actually produced it arrive together, not as two separate acts of trust.
This is the exact problem we set out to address with OxfordAQA, our teaching and learning partner, in delivering the International GCSE Global Skills Projects.
Students work in groups, but they are marked only on their own contribution, never on the group’s combined output. Students and teachers complete candidate record forms authenticating exactly who did what, at every stage, not only at the end. None of it depends on a single test on a single day. All of it depends on evidence collected, and authenticated, as the work happens.
That is a more demanding model for a school to run than setting one final exam. It is also a more honest one, because what it produces is not a claim about a student. It is a record.
This is not a small distinction, and it will not stay a niche one. Schools and universities are, in effect, asking employers to trust what they say about a young person’s capabilities. Employers, on their own evidence, are increasingly unable to take that on trust, from anyone.
Closing the verification gap will not be solved by asking people to trust our assessments more. It will be solved by building assessments that no longer need to be taken on trust, whoever is reading them and whenever they are read.
Because the question was never really whether skills matter. It is whether anyone, on the other side of the transcript, can still tell the difference between a skill and a claim.
