Africa stands out for having relatively high female participation in STEM education, an area that usually dominates conversations about women in technology. Women account for 47% of STEM graduates from African universities, a higher proportion than in Europe, Asia, North America, South America or Oceania. The trouble begins when those graduates try to turn their qualifications into careers.
Their representation falls to between 23% and 30% in technology roles, according to McKinsey research published in 2025, before dropping further as careers become more senior. Fewer than 12% of technology leadership positions in Africa are held by women and only about one in 10 African tech start-ups has a woman CEO. Nearly half of the women emerging from STEM programmes are effectively disappearing somewhere between graduation and the technology workforce.
That makes the familiar description of a “pipeline problem” increasingly inadequate. Africa certainly still needs more girls to have access to science and technology education, particularly in countries and communities where those opportunities remain profoundly unequal. But producing more female STEM graduates won’t fix an industry that struggles to absorb, retain and promote the women it already has.
Masego Mosala, a solutions engineer at Cloudera, thinks too much attention is still being paid to what happens before graduation. “The pipeline exists, but it’s leaking,” she says. “What’s missing is an ecosystem ready to employ and invest in the women who make it through in the first place.”
The implications extend beyond diversity targets. South African companies are simultaneously complaining about shortages of technical skills and trying to accelerate their use of artificial intelligence, even as generative AI adoption is outpacing the skills needed to run it. Pnet says demand for AI-related skills in South African job vacancies has risen 352% since 2019. Losing technically trained people under those conditions isn’t simply an equality problem. It’s an unusually wasteful labour-market problem.
A qualification isn’t the same thing as workplace access
One of the sharpest losses happens almost immediately. McKinsey identified the transition from education into work as one of three points where women’s representation in African technology drops significantly, with its research suggesting that almost half of Africa’s female STEM university graduates aren’t making it into the technology workforce at all.
There are several possible explanations for that gap, and treating all of them as the same problem would be another mistake. Discrimination in recruitment isn’t the same as a lack of entry-level opportunities. Neither is the same as graduating without the practical experience employers expect. Family expectations, economic constraints, geography and access to professional networks can further affect whether someone who earns a STEM qualification eventually works in technology.
Mosala points to another problem she encountered moving from university into the South African job market: the difference between being academically capable and knowing how to function inside an organisation. “Graduates need to code, yes, but they also need to pitch an idea in a meeting, defend a technical decision to a non-technical stakeholder, and work a room.”
Universities can’t carry that responsibility alone. Employers frequently complain that graduates aren’t “work ready”, while expecting universities to anticipate the working practices of an industry whose technologies and skills requirements can change substantially during the time it takes someone to complete a degree.
There is a reasonable argument for more practical and project-based university work, particularly when it exposes students to collaboration, deadlines and problems without clean textbook solutions. There is an equally strong argument that companies need to stop treating graduate development as somebody else’s job.
Internships, learnerships and vocational programmes are supposed to occupy some of the space between education and experienced employment. They become considerably less useful when organisations approach them primarily as compliance exercises or expect graduates to arrive already carrying the judgement that normally comes from years of actually doing the work.
Then there are the rules nobody writes down
Technical ability is only one ingredient in building a technical career. Workplaces contain another layer of knowledge that is rarely written into a job description and tends to become obvious only after someone has spent enough time navigating one. Mosala calls it the “hidden curriculum” of the job market.
It includes knowing how to survive several rounds of technical interviews, how to use LinkedIn as something more than a digital CV, when to advocate for yourself and how to make sure good work is noticed rather than assuming somebody eventually will notice it. None of those abilities appears particularly mysterious once someone explains them, but access to the people who can explain them isn’t evenly distributed.
Professional networks pass knowledge around informally. Someone whose family, university community or social circle already includes people in established careers can learn how recruitment and corporate advancement work long before being formally taught. Others have to discover the same rules after making mistakes, watching colleagues or finding somebody willing to explain them.
The difference becomes more consequential when the networks themselves reflect historical patterns of privilege. If advancement partly depends on knowledge acquired informally, then who gets invited into those informal spaces matters.
That helps explain why simply hiring more women isn’t enough. McKinsey’s survey of 220 women working in technology across African markets found that roughly a quarter reported gender bias and limited opportunities for skills development as barriers to advancement. Only 10% reported encountering no barriers to career progression at all. Among respondents who had left the technology workforce prematurely, half had between six and 10 years of experience, which means the industry isn’t only losing women at the entrance. It’s losing some after they’ve already accumulated years of scarce technical experience.
Mentorship helps, but access is more complicated than mentorship
Technology companies have become comfortable talking about mentorship, and there are good reasons for it. Having someone willing to explain how an organisation works can shorten the period in which younger employees are expected to decode everything themselves.
Mosala’s own experience suggests that mentorship is only part of what useful professional support looks like. She describes having mentors who shared lessons from their careers, sponsors who could create opportunities she couldn’t create for herself and coaching that provided accountability. Those roles aren’t interchangeable, because a mentor can offer advice while a sponsor has enough organisational influence to mention someone’s name when that person isn’t in the room.
That difference becomes particularly important once career advancement depends less on completing assigned work and more on being trusted with larger responsibilities, clients, budgets or teams. Advice can prepare someone for an opportunity, but it can’t guarantee they will ever be considered for it.
This is where well-intentioned programmes can become deceptive. An organisation can have a women’s network, mentorship scheme and annual diversity event while leaving the mechanisms that determine promotion essentially untouched. Counting how many women joined a programme reveals very little if nobody checks what subsequently happened to their careers.
The harder measurement is whether women are being hired, retained, promoted and given access to positions from which they can progress again. A programme can be busy, popular and highly visible without changing any of those outcomes.
We may not know enough about the women who disappear
There is another uncomfortable weakness in the way the industry talks about its talent pipeline: much of the available information comes from people who successfully entered it. Mosala argues that companies should be talking specifically to women who completed STEM qualifications and then never entered technology.
“Ask exactly when they walked away, and why,” she says.
That group could tell us something considerably more useful than another survey establishing that gender inequality exists. A woman who applies repeatedly for technical roles and isn’t hired has encountered a different failure from someone who never applies because the jobs are concentrated somewhere she can’t afford to move. A developer who leaves after eight years because advancement has stalled presents another problem again.
Those distinctions change what an effective intervention looks like. If the problem is recruitment bias, another coding programme won’t solve it. If graduates lack practical exposure, changing promotion policies won’t help them get their first job. If women are leaving after six or 10 years because advancement, workplace culture or family responsibilities make staying unattractive, increasing the number entering university simply feeds more people into the same loss further downstream.
Companies should be capable of treating this with the same appetite for evidence that they claim to bring to product development: define the problem, try an intervention, decide what success should look like, track what happens to different cohorts and change the approach when the results are poor.
Instead, diversity programmes can become strangely resistant to the discipline businesses apply elsewhere. Participation is reported as progress, initiatives survive because they exist, and another annual report arrives without answering the more difficult question of whether women’s careers actually changed.
Africa’s unusually high proportion of female STEM graduates should be an advantage, but instead it exposes how much of the harder work begins after education. Getting women into lecture halls was never going to be enough if the institutions waiting outside them can’t turn those qualifications into access, progression and careers worth staying in.
The continent doesn’t have an unlimited supply of engineers, developers, data scientists and AI specialists to draw from, and South African employers are already competing for people with precisely those kinds of skills. Training women for those careers and then accepting the steady erosion of their representation as they move from university into work and towards leadership isn’t merely a failure to make technology more representative. It means building scarce human capital and then designing workplaces that are remarkably good at wasting it.