Something quietly strange is happening in Africa’s digital economy. The continent is getting steadily better at producing people who can work digitally. It is not getting better at anything like the same rate at producing paid digital work for those people to do. That gap is widening, and it is beginning to matter more than the connectivity gap that dominated the last decade of policy.
For most of that decade, the strategy was to expand supply, and it was the right call. In 2019 the IFC projected that by 2030 roughly 230 million jobs in Sub-Saharan Africa would require digital skills, and put the value of training that workforce at around 130 billion dollars. Later World Bank analysis placed Kenya in a similar bracket, with around half of all jobs expected to require some digital competence by 2030. Set against a tertiary system where, by one Boston Consulting Group estimate, only about a tenth of graduates leave with formal digital training, the case for coding academies and skilling programmes almost wrote itself.
Those investments were necessary and they are producing real capability. But a projection of jobs that could require digital skills is not a guarantee that someone will pay for that work. The 230 million figure is often quoted as a job-creation promise; it is closer to a demand forecast that assumes growth arrives on schedule. A 2020 Brookings paper argued that Africa’s youth employment challenge is better understood as a missing-jobs crisis than a missing-skills one: the binding constraint is a shortage of formal, paying work, not a shortage of willing workers. The World Bank’s 2025 Africa’s Pulse, subtitled Pathways to Job Creation, sharpened the point: around twelve million young Africans enter the labour market each year, and only about three million of them find formal wage jobs. The question is shifting from “can Africans do the work” to “who is going to buy it.”
I want to look at that question through one small window, and I should be transparent about the glass. I founded WorkKE, a Kenyan digital-work marketplace, and the figures that follow describe activity on that single platform in its first year. They are not a measurement of Kenya’s freelance economy, still less of Africa’s, but one operator’s dataset, useful because marketplace data exposes behaviour that surveys tend to miss.
Between September 2025 and August 2026, according to internal figures reported independently by Condia, about 7,140 freelancers registered on the platform against 294 employers and 281 jobs posted. Of those freelancers, 757 submitted at least one proposal, 39 won a bid, and 33 completed paid work. Fewer than five in every thousand people who signed up finished a job. Completed jobs were small: Condia put the median at 2,000 shillings, roughly 15 dollars, with the average pulled up to about 36 dollars by a handful of larger contracts. Cumulatively the platform has moved more than five million shillings to freelancers, a figure separately reported by Disrupt Africa.
The tempting reading of that funnel is the raw ratio: thousands of workers, a few hundred jobs, so Africa simply lacks employers. That reading is not wrong, but it is shallow, and it points at the least interesting part of the data.
The more revealing numbers sit one level down. Of the 294 businesses that registered as employers, only about 105 ever posted a single job. Among those that did, 46 came back for more, and those repeat buyers generated 222 of the 281 jobs. One in six registered employers accounted for close to four in five of all the work. On a young marketplace that is not a failure signal. It suggests that hiring digital help, once it works, tends to repeat. A business that has never contracted anyone online treats the first attempt as a risk; one that has done it once treats the next as routine. The harder problem, then, may not be the crude one of “too few employers.” It may be conversion: how a firm moves from never having hired online, to having hired online once, to treating outsourcing as an ordinary part of how it operates.
Causation deserves caution. Repeat buyers may simply be the larger or more digitally comfortable firms, and a handful of power users can flatter any early dataset. But the pattern fits what marketplace operators see repeatedly: trust is the gate, and repeat purchasing is what sits on the other side of it.
This is where the economics underneath the technology start to matter. Every two-sided marketplace faces a cold-start problem, and its failure mode is badly understood. Adding more workers does not strengthen a thin market; it does the opposite. When supply races ahead of paying demand, the result is crowding: many people competing for few jobs, downward pressure on prices, and eventually churn as newly trained workers conclude there is nothing here for them. The asset that actually matters is liquidity: reasonable confidence that a job posted will be filled and a skill offered hired. It is built on both sides at once. Workers need to be discoverable and credibly vouched for. Buyers need to overcome the fear of being defrauded, the difficulty of judging a stranger’s competence, and plain unfamiliarity with managing someone they will never meet. A platform that solves only the supply side has solved the easier half.
None of this is fixed by importing the Silicon Valley marketplace template wholesale. The Upwork and Fiverr model was built to route developing-country supply toward dollar-denominated Western demand, and much of Africa’s digital-work conversation still treats foreign clients as the only serious source of income. That is worth challenging. Small and medium firms are about 90 percent of businesses and more than half of employment worldwide, on the IFC’s figures, and in Africa the overwhelming majority are micro and informal: seven in ten African workers are self-employed, and fewer than a quarter of the continent’s businesses use advanced digital tools with any intensity. A shop owner does not need a full-time designer, accountant or social-media manager. She needs each of them occasionally. That periodic, low-value, high-trust demand is a genuinely different market from export freelancing, and it is largely domestic: African businesses buying from African professionals.
The encouraging part is that one piece of the infrastructure this market needs already exists. Mobile money moved an estimated 1.4 trillion dollars across Sub-Saharan Africa in 2025, according to the GSMA, about two-thirds of the global total. The rails for moving money are, unusually, not the problem. When the median job is worth 15 dollars, though, transaction costs and dispute handling have to be tiny and fast, or the trade never clears. What remains underbuilt is the trust layer around the payment: identity, reputation, escrow, recourse when work goes wrong, and the quiet employer education that turns a referral-based hiring habit into a digital one. Informal networks are not a backward practice to be swept away. They are the incumbent trust technology, and the task is to digitise the trust they already carry rather than to replace it.
Artificial intelligence complicates this, and the generic version of the point is useless. AI is making digital production cheaper and more abundant, so a freelancer who could produce one unit of work can now produce several. But raising output per worker does nothing on its own for demand. If a hundred people can suddenly do what ten used to, and there are still only ten buyers, the technology has improved supply and made the liquidity problem worse, not better. The more interesting question is whether AI also lowers the cost of being a buyer: helping a small business write a brief, scope a job, compare proposals and supervise a contractor it does not fully understand. That is demand-side leverage, and it is the side almost no one is building for. Which effect dominates is not a property of the technology. It is a choice about what we build.
For policymakers and investors, the implication is a change of scorecard more than of ambition. A digital-work programme measured by people trained will always look successful; measured by income earned and transactions completed, many would look very different, and the second measure is the one that matters. Worker training needs a demand counterpart: support for SME digital adoption, so that firms learn to buy, not only citizens to sell. Trust, identity, escrow and dispute resolution deserve to be treated as digital-work infrastructure, the way roads and payment systems are. And a continent that keeps pointing its digital labour at foreign buyers should ask why so little effort goes into activating the demand inside its own economy.
Africa does not need to stop training digital workers. The skilling was the right first move, and abandoning it would be a mistake. But skills without a functioning market to absorb them produce something worse than idle capacity: they produce trained disappointment. The connectivity has been laid and the money rails are among the best in the world. What sits between them, still mostly unbuilt, is the market itself.
Disclosure: The author is the Founder and CEO of WorkKE. WorkKE data referenced in this article represents activity on the platform and should not be interpreted as representative of Kenya’s or Africa’s digital-work economy.
