Are we measuring employability effectively?

Skills England's new employability framework describes what's needed to start a career, HESA's survey counts how many graduates have started their career, but neither shows what people went through to get there.

Skills England has just launched an official picture of employability. Thirteen skills, each described for a first job, with prompts for how employers can assess and develop them. It's careful, practical and long overdue.

But does it provide an answer to "what actually is employability?". Put it next to HESA's Graduate Outcomes survey and you have a dictionary and a thermometer. One tells you what skills should look like on day one of a career. The other tells you what happened fifteen months after graduating. Neither is a spirit level, and a spirit level is what you'd need to find out whether the table is flat.

Skills England has defined the first day, not the career

The framework sets out thirteen skills, mapped to the Standard Skills Classification, from planning and organising through to writing. Each is defined at the start of a career, with what it looks like in a first job, how a young person might have developed it, how employers can assess and build it, and what it means in an AI-enabled workplace.

It exists for a good reason. Skills England's own youth employability work found that skills young people build through sport, caring, volunteering and lived experience are largely invisible to the people hiring them, and that employers and young people have no shared language. That's a real problem, and it isn't a new one. Youth Employment UK reviewed 18 frameworks and 86 reports in 2017 to argue for a unified set. Tushar and Sooraksa's review found 87 unique skills across 25 studies spanning thirty years, with the same handful of skills on top every decade. The vocabulary was never the gap. Recognition was.

The AI columns are quietly clever. For nearly every skill they say the same thing: use the tool, check the output, own the result. What's being asked for is judgement, not tool fluency.

All of this makes it a useful job-readiness framework. Clark and colleagues define a work-ready person as one with the foundational skills to be minimally qualified for a specific occupation. Readiness asks whether you can start on Monday. Employability asks whether you'll still be worth hiring in ten years.

As a definition of employability, it's thin. There's no career identity, no networks, no health or economic capital, no labour market. The American NACE competencies include career self-development and networking. The English version of thirteen skills doesn't.

Capital theory shows what's missing

Employability research changed shape around 2015. Donald, Baruch and Ashleigh's 2023 review traces the move from skills checklists to capital-based models, and pulls them together into nine forms of capital: social, cultural, psychological, personal identity, health, scholastic, market-value, career identity and economic. Skills live mostly inside two of the nine. The rest is who you know, what you could afford, how you handle setbacks, and whether you can tell a story about where you're heading.

Job readiness sits inside this model, filed under market-value capital; industry-driven technical and personal skills, work experience, placements. It's one layer of nine. A 2019 paper in Education Sciences & Society noted that work readiness has no agreed definition and is, to some extent, overlapping with and replacing employability. Treat it as a layer and the confusion goes away. Treat it as a synonym and a lot of context gets lost.

The model also treats things a graduate didn't choose, including social class and parental financial support, as part of employability rather than noise around it. And it sets a separate list of external factors beside the capitals: labour market demand, recruitment bias, chance events, cost of living against wages. Capitals are things a person can build. External factors are things that are done to them.

Critical employability scholars have been making a version of this point for years, that the discourse overstates individual responsibility and underplays social position. The capital model is one of the few frames that puts both halves in one place. If a definition is meant to explain who gets hired, it needs both. It also allows a young person to understand all the contributing factors to the career challenges they may face, rather than them feeling that it's purely down to their own skills gap.

The HESA survey counts the aftermath

The latest release covers the class of 2023/24, just over a million graduates surveyed, 353,755 responses. HESA reports 87% in some form of work or further study, 57% in full-time employment and 7% unemployed. That's an outcome. Employability is a capacity, and an outcome is capacity multiplied by demand, bias and luck. Brown and colleagues tested the capacity side using a large Australian graduate survey and found graduates' self-rated skills barely moved the odds of being employed, which is why they warned against building institutional performance indicators on such measures.

HESA hasn't ignored quality. It publishes a "highly skilled" share and asks three reflective questions: is your activity meaningful (85%), does it fit your future plans (76%), are you using what you learnt (68%). However, these questions can be left to interpretation as terms such as 'meaningful' can be fairly subjective.

The privilege gap is visible, but only as a label

HESA's data can be cut by background. Jisc's dashboards split outcomes by ethnicity, disability, sex, age, POLAR4 and the deprivation indices. Wonkhe's analysis of an earlier release found English graduates from the most deprived quintile were less likely to be in paid employment, and Graduate Futures Institute analysis of 2022/23 data has disabled graduates at 79% employment against 83%. So the gaps are visible. But they sit in dashboards, not headlines, and they rest on area-based proxies. POLAR describes where someone lived, not their household.

Privilege is harder to see. Wonkhe's reading of the Department for Education's LEO data noted that the group with unknown 'free school meal' status had the highest earnings, suggesting that it may be that they were privately schooled as the reason. If so, disadvantage gets a flag and privilege gets a blank.

The gap itself is well documented. The IFS found graduates from the richest fifth of households earn around 10% more a decade on, with the same university and subject. A 2025 UCL and Nuffield Foundation study found working-class graduates 32% less likely to get an offer from large employers. None of that is a skills gap.

Contextual recruitment tackles this at selection. It reads an application against background markers, which is a correction applied to a label. Labels are often looked at from a segmented perspective without any acknowledgment of intersectionality. Basing assumptions on labels alone can be misleading as the context doesn't go deep enough.

A baseline isn't an explanation

A recent HEPI post by Leeds University Business School academics makes the warning plainly. When demographic characteristics become the main lens, responsibility shifts from systems onto the groups, and data meant to expose inequity can end up reinforcing a deficit view. Their answer is to ask which institutional conditions produce unequal outcomes, which is important. The practical question is what you'd measure to find out.

Start with what labels can't do. They hide variation. The same Graduate Futures Institute analysis that put disabled graduates at 79% employment found autistic graduates at 72%, with 12% unemployed. Wonkhe noted of the LEO release that the only intersection available was with gender. A group average is an average of very different lived experiences.

They can't separate causes that need different fixes. Where research does look at experience, it finds real differences. Sutton Trust polling found 46% of disadvantaged pupils take part in extracurricular activities against 66% of better-off peers. The Education Policy Institute found pupils on free school meals less likely to attend sports and arts clubs, with independent school pupils far more likely, and linked club attendance to higher odds of university progression and employment in the early twenties.

The Graduate Futures Institute and University of Leicester's new report names networks, work experience, employer knowledge and confidence as unevenly distributed. If a gap comes from missing enrichment, that's a school and university problem. If it comes from filtering at application, that's an employer problem. If it comes from not being able to afford unpaid experience, that's a financial problem. A demographic gap lumps all three together. It invites the lazy thinking that the group is the problem, which is exactly what the HEPI authors warn against.

Skills England's framework shows the same blind spot. However, its wider work recognises that skills come from informal routes like caring, and that young people who are NEET are a mix of circumstances rather than one group. The framework's "how a young person might have developed it" column lists caring responsibilities, part-time work and family responsibilities, and its written guidance rewards clarity over polish. But the same column leans on clubs, team sports, residential and outdoor experiences, and enrichment, exactly where access differs, and the framework document doesn't say how to weigh a caring role against a Gold expedition. Its assessment advice asks candidates to talk through examples, show confidence and be noticed for how they treat the interviewer. A shared language could end up rewarding the people who already speak it.

Keep the baseline, then ask what it's made of

None of this means dropping what's already out there. Comparing outcomes against a starting point is a good baseline, but outcome statistics should carry individual-level background (free school meals, first-generation status, school type) rather than only area proxies, so privilege becomes a category rather than a blank. A baseline tells you someone is further behind. It doesn't tell you why.

How could we improve on the current systems in place?

First, capture experience alongside demographics. Ask what people had access to and had to do as part of their experience; enrichment, work experience, who they could call for career advice, paid work or caring during study, how their living costs were funded. These are the things the research already ties to outcomes, from the capital model's economic and market-value forms to the Graduate Futures Institute's findings. Most are associations rather than proven causes, which is why they need measuring properly. And the data should describe the conditions people met, not the deficits they carry. That's the line the HEPI authors are drawing, and the design has to hold it.

Second, measure the behaviours that experience builds or doesn't, such as how people build networks and seek out opportunities, and hand the result to the individual rather than the league table. Brown and colleagues' caution points the same way. Self-assessed measures are weak performance indicators but may still be useful development feedback. Access and outcomes are well covered in research and data, but the behaviours in between mostly aren't.

Third, enable young people to have access to data along their journey so they can distinguish between the challenges they face and the factors that they had less control on but may have influenced their outcomes. Being able to understand how to use these differences to their advantage is important. The likelihood is they've had to overcome obstacles they may not be aware of, which have developed resilience, adaptability and perseverance.

Meeting the gap between educators and employers

Graduate outcomes now feed the regulator's conditions of registration, which judge institutions on their graduates' employment record. But a gap that comes from employer filtering isn't a university problem, and a gap that comes from the cost of unpaid experience isn't either.

The capital model already adds an additional layer that's currently being missed, but could help close the gap. It lists actions for students, educators, careers professionals and employers, and for employers that includes equitable hiring processes and awareness of bias in recruitment. What's missing is a measure that tells you exactly what the employability gap is that you're looking at.

Employability is what a person brings, what they can demonstrate, and what the system lets through. The framework describes the middle. The survey counts the aftermath. Nobody is measuring the difference between what a person can build and what external factors have affected them.

So here's the question. If employability is going to be fair, what would a measure look like that separates the two halves of the gap, and who should be answerable for each?