Most Learning Happens at Work, No Certificate Follows

Research consistently shows that 70–80% of job-relevant skill development occurs through informal and workplace learning — on-the-job experience, mentoring, self-directed study, community practice — rather than through formal education. Yet national qualification frameworks in virtually every country only recognize formal learning (degrees, diplomas, certificates issued by accredited institutions). Workers who have developed significant expertise through years of practice — experienced welders, self-taught software developers, traditional construction craftspeople, health community workers — cannot convert their competencies into recognized credentials without completing formal programs that largely re-teach what they already know. This creates a structural barrier to career advancement, mobility, and equitable access to further education.

The ILO estimates that 1.4 billion workers globally are in jobs where formal qualifications do not match their actual competency — mostly workers with competencies above their formal credential level. In the EU, CEDEFOP found that 40% of adults have skills acquired informally that are not recognized by any qualification. The economic cost includes: worker underemployment (skilled workers in lower-paying jobs because they lack credentials), employer screening inefficiency (using credentials as proxies when direct competency assessment would be better), and educational waste (forcing experienced workers to sit through coursework they've already mastered). For workers in the Global South, where formal education access is limited but informal skill transmission is culturally embedded, the credential gap is especially acute.

Recognition of Prior Learning (RPL) systems exist in many countries (France's VAE, Australia's RPL, South Africa's RPL framework, UK's APEL). Implementation data is discouraging: France's VAE, the most mature system, processes only 30,000 validations annually despite 28 million employed workers. Australia's RPL is underused because the assessment process is nearly as burdensome as completing the formal course. South Africa's RPL framework was designed in 2002 but remains largely unimplemented due to lack of assessor capacity. The barriers are consistent across countries: (1) the assessment is expensive (requiring individual portfolio review by qualified assessors); (2) the evidence requirements are unclear (what counts as evidence of competency?); (3) academic institutions resist RPL because it threatens enrollment-based funding models; (4) the process rewards workers who are articulate and document-savvy over those who are most competent. The fundamental mismatch is that informal learning is experiential and context-embedded, while credential assessment frameworks are designed for standardized, decontextualized evaluation.

Scalable, valid competency assessment that does not require portfolio compilation or assessor interpretation — instead, directly assessing what the worker can do. Promising approaches include: (1) digital performance assessment where workers complete realistic work tasks in simulated or real environments, with automated scoring of observable outcomes; (2) AI-assisted evidence gathering that mines workers' existing digital footprints (code repositories, work products, supervisor feedback, client reviews) to compile competency evidence automatically; (3) modular micro-credentials that allow partial recognition — credentialing specific competencies rather than entire qualifications, allowing workers to "fill in" only the gaps between their informal learning and the formal requirement. The success of coding bootcamp placement rates (demonstrating that competency-based assessment works for software engineering) suggests the model is feasible — the challenge is extending it to less digitally documented skill domains.

A team could design and pilot a performance-based competency assessment for a specific trade skill (e.g., welding, electrical installation, community health work) that takes less than 2 hours and produces a reliable competency estimate, comparing results against formal qualification status. A data science team could build an automated portfolio compiler that extracts competency evidence from a worker's digital footprint (GitHub, LinkedIn, review platforms) and maps it against a specific qualification framework. Relevant disciplines: educational assessment, psychometrics, human-computer interaction, labor economics, instructional design.