In early June, I had the privilege of joining a panel convened by the World Bank Group’s High-Level Advisory Council on Jobs in Cape Town, alongside Ravi Naidoo of Youth Employment Service (YES) and Sharmi Surianarain of Harambee Youth Employment Accelerator, in a conversation moderated by Nicola Galombik. The session, “From Skills to Jobs: What Actually Works?” formed part of the World Bank Group’s high-level Jobs Council, co-chaired by President Tharman Shanmugaratnam of Singapore and Michelle Bachelet, former President of Chile. The Council brings together a select group of global leaders from government, business and academia to help the World Bank Group identify and scale solutions for job creation.
Our session grappled with a question that sits at the centre of everything we do at Collective X: how do you translate skills development into employment at scale?
It’s a question everyone in that room is wrestling with. Economies grow and create jobs, but youth employment doesn’t always keep pace. The instinct is often to treat this as a training problem; train more people, and the jobs will follow. Our experience building Collective X tells a different story. The harder, more useful question isn’t what needs to happen. Everyone agrees on that: more young people into digital jobs, employers getting the skills they need, and government backing the pipeline. The harder question is how we actually make it happen, how do we shift a broken market, not just describe it?
Collective X is an answer to the “how”
Launched in 2023, Collective X is a not-for-profit ecosystem orchestrator. We were established to coordinate South Africa’s national digital skills strategy for entry-level ICT roles. We are not a training provider, and we are not an employer. We are the architecture that holds the system together. Our role is to be the trusted intermediary that sits between government, employers and young people, absorbing the tension that change of this kind inevitably creates.
Our approach is demand-led. We start with employers and ask what roles they need filled, now and as the market shifts. Then we align training delivery, catalytic financing and outcomes accountability around it. That sequencing matters. Train for yesterday’s job market, and you produce graduates for roles that no longer exist.
Three phases, each with its own hard lesson
We didn’t arrive at this model by design. We got here through three distinct phases, and each one taught us something that the next phase depended on.
The first was about mobilisation: being formally recognised as a national intermediary by The Presidency and BUSA, and assembling a board of credible business leaders rather than the usual NGO-sector profile. The hard part wasn’t the strategy; it was convincing an ecosystem to trust a body with no track record. Credibility had to be earned before anything could be built.
The second phase was piloting ideas and early implementation. We secured government funding, aggregated multiple donors into a single fund, and tested our model with multiple iterations to build employer proof points. We learned hard lessons about what worked and what needed to change for scale, and implemented those changes. Getting employers to be first-movers, and encouraging government to pay for outcomes rather than inputs was, and remains, genuinely difficult. We also made a deliberate choice here that turned out to matter as much as any technical decision: we had several system-level priorities we could have chased, and we chose to focus on two. Knowing what not to do was as important as knowing what to do.
We’re now in the third phase, early scale-up. We’ve diversified across sectors, built the technology to run an outcomes fund at scale, and secured government commitment for 2026. But I want to be direct about where we are right now: two dependencies remain unsolved. Multi-year government financing, because annual funding cycles can’t support the training pipelines employers actually need. And policy recognition of non-accredited programmes, because right now businesses can’t route their skills spend directly into the roles that matter most. Solve those two things, and this model scales at a completely different order of magnitude.
Five pieces of architecture that aren’t unique to digital skills
What I tried to leave the Council with wasn’t a description of our programme; it was the underlying architecture, because I believe it’s repeatable beyond digital skills and well beyond South Africa.
Business confidence. Employers need proof before they’ll commit, not promises. Someone has to absorb the early risk and demonstrate that trained candidates become productive employees. That’s not a one-off sell; it’s daily, ongoing work.
Government buy-in. Real co-investment, not just endorsement. The government has to be willing to fund outcomes rather than compliance, which takes policy courage and new procurement design.
A capacitated intermediary. This is the most underestimated piece of the puzzle, and the hardest to fund. Donors and government rarely fund the operating capacity of the body that convenes, coordinates and course-corrects, yet without it, nothing else holds together.
Demand intelligence. Granular, real-time data on what jobs exist and how the skills they require are shifting, including under the pressure of AI. Without this, you train for a market that’s already gone.
Blended, scalable outcomes financing. A fund that pools public, private and donor capital and pays only on verified employment outcomes. Simple in principle; genuinely hard to make scalable in practice, and it requires its own change management with donors.
These five pieces are interdependent. You can’t have outcomes-based financing without government buy-in. You can’t build employer confidence without demand intelligence. And none of it holds together without the intermediary, which is precisely the piece most likely to be left underinvested, and the piece that will quietly determine whether a model like this succeeds or fails.
The honest version of what’s been hard
Employers are rational, and rational employers don’t want to be first. Getting the government to fund differently, to pay for retained employment rather than inputs, runs up against systems built for compliance, not outcomes. Change management at this scale never really stops; every stakeholder, from training providers to SMEs, has to behave differently from how they used to. And the intermediary function itself can’t simply be copied. You can replicate a fund structure or an outcomes contract. You cannot template the capability to hold the tension between government timelines, employer needs and youth aspirations. That has to be built, resourced and protected deliberately.
Where this is going
We have a national throughput target of 40,000 youth into in-demand digital roles by 2028, with an ambition of 500,000 by 2033. That isn’t a slogan. It’s a function of unlocking the two policy dependencies I described, and of this architecture being adopted beyond digital skills into other high-growth sectors. The model is proven. What remains is whether we, and others facing the same jobs challenge, have the will to implement it at the scale the problem actually demands.
One question that came up repeatedly during the session, and that I’d add here for anyone thinking about the future of this work: what about AI? Our view is that AI is reshaping the roles we train for. Some entry-level tasks will be automated away, but new roles are emerging that need human judgment, context and communication. A demand-led model is built precisely for this kind of shift. We’re not training for static job descriptions; we’re training for live employer demand. With our soon-to-be-launched LinkedIn-partnered Skills Observatory, we have real-time intelligence on demand and real-time signals on how that’s changing. The organisations that get caught off guard will be the ones still training for yesterday’s market.
It was a privilege to share this conversation with the World Bank Group, and fellow participants who are tackling the same problem from different angles. South Africa doesn’t have all the answers yet, but I left Cape Town more convinced than ever that the architecture we’ve built is one worth sharing.
Evan Jones, Chief Executive Officer, Collective X
