Before you close the Skills gap, make sure it exists.

Leo, a serious business student, has just completed a market-entry analysis for a regional retailer considering expansion into a new GCC market.

Customer research suggested demand, competitor activity raised doubts, and the financial forecast changed sharply depending on which assumptions he accepted.

Asked what he actually did, Leo explains:

LeoIn his words

“The sales numbers pointed one way, the customer interviews pointed another, and I had to decide whether the expansion was worth the risk.”

Dr Chen, the module leader, records the intended learning in academic language:

Dr ChenLearning outcome

Critically evaluate conflicting evidence and justify a strategic recommendation.

Marcus, a head of commercial strategy recruiting graduate analysts, frames the workplace requirement differently:

MarcusJob requirement

Assess market attractiveness and commercial viability to support go/no-go investment decisions.

The three descriptions come from different purposes. Leo remembers the situation he had to resolve. Dr Chen abstracts that experience into the reasoning the module was designed to develop, while Marcus cares about the commercial judgement a graduate analyst must support. Their vocabulary differs enough to defeat simple keyword matching even though the underlying capability is closely related.

Several maps are describing the same capability landscape with different legends. Once those legends become the basis for comparison, a plausible match can disappear into the record as a skills gap.

What exactly is the “skills gap” being asked to explain?

“Skills gap” is often used as though it names one clearly bounded problem. In practice, the phrase covers several diagnoses that lead to very different decisions.

It can refer to a genuine shortage of people with a required capability, a mismatch between workers and the work they perform, a difference in qualification level or field, weak development, thin assessment evidence, unrealistic recruitment requirements or related capabilities that have never been connected because each context describes them differently.

Research on skills mismatch has repeatedly shown why the measurement method matters. Qualification measures, employer reports, worker reports and job analyses observe different objects and can therefore produce different diagnoses (McGuinness, Pouliakas and Redmond, 2018). Cappelli’s review of evidence from the United States likewise showed that hiring difficulty and employer complaints do not, by themselves, establish an economy-wide shortage (Cappelli, 2015). Reviews of Industry 4.0 research continue to report disagreement over how skills gaps are defined and measured, which makes apparently precise gap claims difficult to compare unless the underlying definition and method are visible (Rikala et al., 2024).

The intervention changes with the diagnosis. A genuine shortage may justify development or recruitment. Thin evidence can make existing capability difficult to recognise and sends attention back towards assessment. Poor job design or unattractive conditions can create hiring difficulty that training will never repair. Where the capability already exists under another description, prescribing more development wastes effort because the failure occurred in comparison rather than capability formation.

A skills-gap claim therefore carries an analytical burden. Before anyone redesigns a curriculum, funds a new training programme or concludes that graduates lack what employers need, the institution has to establish what is actually missing and how that absence was identified.

The same capability leaves different records behind

Dr Chen’s learning outcome has a legitimate educational purpose. Learning outcomes help institutions organise curricula, connect teaching with assessment and state what a learner is expected to know, understand or be able to do. Cedefop treats them as an important reference point between education and the labour market while also emphasising that their usefulness depends on how they are formulated and applied in practice (Cedefop, 2022).

Marcus is solving another problem. His job description needs to communicate the work, the conditions surrounding it and the level at which performance becomes valuable to the organisation. The document is written for recruitment and performance expectations, so the educational history behind that capability is largely irrelevant to its immediate purpose.

Leo carries the experience in a less formal record. He remembers the market reports, the forecast he challenged, the assumption he abandoned, the recommendation he revised and the presentation in which he had to defend the final decision. The work remains vivid because he experienced it as a sequence of choices under uncertainty. Months later, however, the analytical judgement embedded in those choices is much harder to name than the assignment itself.

Research has repeatedly identified this disconnect. Rook and Sloan (2021) found different and sometimes competing understandings of graduate attributes and employability among academics, career advisers, professionals and students in work-integrated learning, with many student participants showing limited awareness of the attributes attached to their courses. Swingler et al. (2019) similarly found that students could recognise transferable skills in their education but wanted more help understanding where those skills were embedded, how they were evidenced across contexts and how they connected with employer expectations.

A university can design a demanding experience, attach learning outcomes to it and assess the resulting work while the capability record still becomes progressively thinner. Dr Chen’s outcome remains in the course documentation. The final grade may combine several dimensions of performance and reveal little about Leo’s individual reasoning. By the time Marcus sees an application, the original conditions, judgement and evidence may have collapsed into a module title or a broad skills claim.

This thinning matters because the assignment itself is an unreliable proxy for capability. Two students can complete the same market-entry report and receive the same grade while demonstrating materially different performances. One may interrogate contradictory evidence, revise an early assumption and defend the consequences of that revision. Another may follow the expected structure competently without making the same analytical move. “Completed a market-entry report” preserves the container while discarding much of what would allow another person to judge the quality of the performance inside it, so a useful capability record has to retain enough of that performance to support later interpretation.

False difference and false sameness

Dr Chen expects Leo to “critically evaluate conflicting evidence and justify a strategic recommendation”. Marcus wants a graduate who can “assess market attractiveness and commercial viability to support go/no-go investment decisions”.

Their sentences belong to different professional languages, yet both require the person to examine competing information, judge its relative weight and reach a defensible conclusion under uncertainty.

HELFFE calls the failure to recognise that relationship false difference. Related capability meanings appear separate because each context encodes them in its own vocabulary. A comparison built mainly on surface wording can therefore exaggerate distance between education and work even when the underlying performance has substantial overlap.

The familiar label “critical thinking” creates another problem. Dr Chen may place it in a learning outcome, Marcus may add it to a job description and Leo may write it on his profile. The repeated term creates an appearance of alignment before anyone has established what performance each actor expects. In one context, critical thinking may mean checking the reliability of a source. Elsewhere it may involve testing assumptions, comparing competing explanations, challenging a commercial claim or making a decision from incomplete evidence. HELFFE treats this second failure as false sameness: identical labels conceal differences in performance, context or expected level.

The conceptual literature gives good reason to take both failures seriously. Definitions of competence have developed through different national and disciplinary traditions, combining knowledge, skills, behaviours and attributes in different ways (Le Deist and Winterton, 2005). Jones (2009) also showed how apparently generic attributes such as communication and problem solving are interpreted through disciplinary cultures and can remain implicit in practice.

Language therefore influences the diagnosis before any development decision is made. Different words can conceal a credible relationship, while familiar shared labels can create agreement that disappears as soon as the required performance is examined.

Recruitment often discovers the translation problem too late

When Marcus invites Leo to interview, Leo receives familiar advice: articulate your value clearly and prepare a strong example.

The advice is sensible, but the institution has already allowed most of the evidence required for that articulation to scatter across module descriptions, old submissions, feedback comments and memory.

Leo now has to reconstruct several years of learning under recruitment pressure. He needs to remember what he actually did, distinguish his own performance from the assignment brief, identify which parts matter to the role and translate an educational experience into language Marcus can interpret. The capability may be strong, yet the record supporting it was never designed to make that translation easy.

Research on transfer makes automatic portability an unsafe assumption. Jackson’s study of 674 business graduates from 39 Australian universities found that the transfer of graduate skills into the workplace was shaped by characteristics of the learner, the programme and the workplace (Jackson, 2016). The meaning and usefulness of a capability therefore depend partly on the conditions in which it is mobilised, which is precisely why a generic label carries so little information on its own.

Skill articulation also responds to deliberate development. Tomasson Goodwin et al. (2019) examined 3,400 students across 44 courses in Canada and found that prior exposure to a structured method for explaining a situation, the action taken and the result achieved was the significant factor associated with later skill-articulation performance. Simply progressing through university did not produce the same association.

Leo still has work to do. Interpreting experience and communicating capability are legitimate parts of professional development. The avoidable burden appears when the student is expected to reconstruct a complete capability record from an educational trail that preserved the assignment more carefully than the performance.

A richer record starts while Dr Chen can still observe the work and Leo can still explain the decisions behind it. It can preserve what Leo did, which knowledge informed the action, how his judgement changed after feedback or failed assumptions, how demanding the context was and what evidence supports the claim. Later translation becomes more credible because the capability has retained enough structure to survive beyond the module.

A common language has to preserve the differences that matter

The Unified Skills Map provides a shared structure through which educational, student and employer descriptions can be understood and connected without forcing every actor into one vocabulary.

Dr Chen can retain language appropriate to teaching and assessment, Marcus can describe the work in commercial terms, and Leo can explain the experience in language that remains natural to him. The common structure sits underneath those descriptions and supports comparison at the level of capability meaning.

That shared layer is useful only when it preserves the distinctions that affect judgement. Context, proficiency and evidence can materially change what a claim supports even when the capability label appears stable.

The expression “can present” illustrates the problem. Leo may have delivered a familiar update to classmates after extensive preparation; Marcus may require a graduate analyst to defend an uncertain recommendation before senior decision-makers who challenge the assumptions in real time. Both performances involve presenting, but the audience, stakes, autonomy and expected standard produce different evidence about capability.

A practical common language should therefore make related descriptions easier to connect while keeping those consequential differences visible. Translation becomes useful when it reduces accidental variation in wording without flattening the conditions under which performance occurred.

The European Commission’s work linking qualification learning outcomes with ESCO skills shows why semantic connection cannot be delegated to surface wording alone. The process combines automated suggestions with human verification because educational and occupational descriptions may be closely related without making that relationship self-evident (European Commission, n.d.). The need for verification also protects against the opposite error: a lexical match can look convincing while masking important differences in meaning.

For HELFFE, the value of a common skills language lies in the quality of the comparison it enables. It should help institutions see relationships that vocabulary obscures, while preserving enough context to prevent a neat map from becoming a misleading one.

A better diagnosis changes the intervention

Clearer capability language improves the point at which institutions decide what kind of problem they are facing.

Where capability is genuinely absent, development may be justified. A mismatch in proficiency can instead require more demanding practice or a different performance context, while weak evidence directs attention towards assessment. Problems in signalling concern the way credible capability information travels and is interpreted. When two descriptions already refer to substantially related capability, translation can recover the connection before another training intervention is added.

Shared language has defined limits. It cannot develop Leo’s analytical judgement, strengthen weak evidence after the fact, guarantee fair recognition, compensate for poor job design or remove the need for contextual judgement.

Its contribution is diagnostic: it makes comparison more disciplined and reduces the chance that vocabulary decides the outcome before the underlying capability has been examined.

That discipline matters for every actor in Leo’s story. Dr Chen needs to know whether an employer requirement points to a genuine development need before redesigning the curriculum. Marcus needs enough precision to distinguish missing capability from weak signalling or unrealistic recruitment expectations. Leo needs a record that allows him to connect what he has actually done with what a professional context is asking him to do.

The three records should therefore be read against a common legend before the blank space between them is called a gap. Only then can an institution determine whether the problem lies in capability development, proficiency, evidence, signalling, job design or translation, and choose an intervention that addresses the problem that actually exists.