Digital skills — the real bottleneck of SME transformation.
The Future of Jobs Report 2025 places skill gaps first among the obstacles to business transformation in 52 of the 55 economies it covers. The OECD’s SME evidence narrows the diagnosis: what separates firms is less how much they learn than whether the learning is organised.
Gli Insights sono pubblicati in inglese.
The World Economic Forum published the fifth edition of its Future of Jobs Report in January 2025, drawing on more than 1,000 employers representing over 14 million workers across 22 industry clusters and 55 economies. Asked which barriers they expect to hinder their transformation between 2025 and 2030, respondents gave the same first answer almost everywhere: skills gaps in the labour market, cited by 63%, ranking first in 52 of the 55 economies and in 19 of the 22 sectors.
Other barriers in the same figure rank lower: organisational culture at 46%, an outdated regulatory framework at 39%, inadequate data and technical infrastructure at 32%, a shortage of investment capital at 26%.
The size of the requirement is arithmetic, and the first sign that training reduces it is now measurable
Employers expect 39% of workers’ core skills to change by 2030, down from 44% in the 2023 edition and 57% in 2020. The Forum offers an explanation it is careful not to overstate: one element contributing to the decline may be “a growing focus on continuous learning, upskilling and reskilling programmes”. Alongside it, the share of the workforce reported as having completed training under employers’ learning and development strategies rose from 41% in 2023 to 50% in 2025, across almost all industries.
Of a representative group of 100 workers, employers expect that by 2030 forty-one will not need significant training, twenty-nine will be upskilled in their current role, nineteen will be upskilled and redeployed elsewhere, and eleven will need training they are unlikely to receive. Fifty-nine in every hundred therefore need training within five years, and close to one in five of those is expected to go without it.
Employers intend to act, and to pay: upskilling is the most common workforce strategy for the period, planned by 85%, ahead of accelerating automation (73%) and hiring for new skills (70%), and 86% expect to fund it themselves.
In smaller firms the binding constraint is the calendar, and the skills channel is informal
The OECD’s Digital for SMEs survey, published in April 2026, supplies the small-firm view: 2,018 responses across twelve OECD countries, gathered through the digital platforms respondents trade on. The OECD states that the sample is neither randomised nor representative; 92% of respondents are self-employed, micro or small businesses.
Asked which barriers they meet when using digital tools, respondents put maintenance costs first at 39%, then lack of time for training at 38%, hardware costs at 37% and the cost of training at 23%. Two of the four leading obstacles concern training, and both point at the same object: the hours in which learning happens.
How those firms obtain capability is the second finding. Asked how they address their digital skills needs, respondents name an internet search (32%), external consultants or experts (29%), friends and family (24%) and generative AI (21%); the OECD describes these pathways as informal and fragmented. Intent is not what is missing: 70% of AI users describe an approach to integrating AI rather than none, and 60% of respondents describe a digitalisation process rather than none.
Regression work on the largest national sub-sample — 1,376 Japanese respondents — tests whether the channel matters. Firms addressing skills needs through structured programmes with specialist institutions, or through internal capacity building such as internal training and mutual learning, are more likely to reach higher digital maturity; informal approaches show no clear association. The OECD is explicit that these are conditional associations, not causal effects, on a sample that is not representative. Read with that caution, the finding still points somewhere useful: what separates firms is less how much they learn than whether the learning is organised.
European statistics show a training gap that widens with firm size rather than closing
Eurostat measures the behaviour itself. In 2024, 22.3% of EU enterprises with ten or more employees provided training to develop or upgrade the ICT skills of their personnel: 17.2% among those with 10 to 49 employees, 41.4% among those with 50 to 249, and 72.6% among those with 250 or more.
The direction is not convergence: between 2020 and 2024 the small-enterprise rate rose 2.1 percentage points from 15.1%, the large-enterprise rate 4.8 points from 67.9%, widening the distance from 52.8 to 55.4 points.
The 2025 wave of the same survey asked enterprises that had considered AI technologies but not adopted them what stopped them. A lack of relevant expertise was the most cited reason, at 70.3%, ahead of unclear legal consequences (53.6%), data-protection concerns (52.7%) and cost (38.4%). Expertise outranks cost by close to two to one, and barely varies with size: 70.9% among enterprises with 10 to 49 employees, 65.1% among those with 250 or more.
Switzerland enters the period with above-average skill disruption and a training system that pays individuals
Employers surveyed in Switzerland expect 41% of core skills to change by 2030, above the 39% global average. Ninety-six per cent expect AI and information-processing technologies to transform their operations, and 73% intend to complement and augment their workforce with new technologies. Retention is the distinctly Swiss concern: 36% expect talent retention to worsen, close to twice the global share.
The Swiss training base is broad, but organised around the individual. The Federal Statistical Office’s Microcensus on Education and Training found that 45% of residents aged 25 to 74 took part in continuing education in the twelve months before the 2021 survey, and 54% of those in employment; computing was the third most attended work-related subject area, at 17%. The office attributes the fall from 62% in 2016 largely to pandemic interruptions of in-person teaching, and notes that design changes leave the two years only partly comparable.
One Swiss instrument is easy to overlook because it is paid to people rather than companies: the Confederation reimburses 50% of eligible course fees for courses preparing for federal examinations, up to CHF 9,500 for a Berufsprüfung and CHF 10,500 for a höhere Fachprüfung. The State Secretariat for Education, Research and Innovation pays the candidate directly, once the examination has been sat, whether or not it is passed. Awareness is the recurring obstacle: among OECD respondents that had not benefited from any public digitalisation programme, 65% said they did not know such measures existed.
Nothing in this evidence calls for a new budget line; it calls for a scheduling decision
These are ordinary management decisions, available to a company of ten people as much as to one of a thousand.
- Put learning in the calendar before putting it in the budget: lack of time is cited by 38% of the OECD’s respondents, the cost of training by 23%.
- Prefer organised learning to searching. Structured programmes and internal capacity building are associated with higher digital maturity in the OECD’s Japanese sample; ad hoc online resources are not.
- Attach the learning to one named process — quotation, scheduling, service response — rather than to a tool. That gives the training a test it can pass or fail.
- Track the number employers themselves report: the share of the workforce that has completed training under a learning and development strategy. The global figure moved from 41% to 50% between the two editions.
- Check what has already been appropriated. In Switzerland the federal contribution to preparatory courses is claimed by the employee, not the employer.
The next observations arrive on a known calendar
Three series will show whether the gap narrows: the Swiss Microcensus on Education and Training, in the field during 2026, the first post-pandemic reading of Swiss participation; Eurostat’s enterprise ICT-training indicator, collected every second year since 2020 and last reported for 2024; and the OECD’s D4SME survey, run annually.
One obligation runs alongside them. The Artificial Intelligence Act, Regulation (EU) 2024/1689, requires providers and deployers to take measures supporting AI literacy among staff and others operating systems on their behalf — an article rewritten by the Digital Omnibus on AI, Regulation (EU) 2026/1744, in force since 27 July 2026, which removed any requirement to guarantee a particular level for any individual. Because the Act reaches deployers outside the Union where the system’s output is used inside it, a Swiss firm whose AI-assisted output reaches an EU market is within its scope. The obligation is expressed as effort rather than outcome, which places it where the rest of this evidence sits: in what a company schedules, and for whom.