Across Sub-Saharan Africa, women run an estimated 58% of informal businesses. They’re the backbone of local food systems, household economies, and community resilience.
Yet when development organisations design coaching and advisory programmes for this group, they routinely rely on tools that women can’t use: smartphone apps, online platforms, WhatsApp groups.
Millions of women-led businesses go unsupported not because coaching doesn’t work, but because the tools designed to deliver it were never built for them.
Training alone doesn’t lead to better outcomes
The evidence on entrepreneurship training points to more varied impacts than often assumed. Reviews of business training programmes consistently find that while they improve knowledge and some practices, their effect on profits and income is less consistent.
McKenzie and Woodruff’s widely cited synthesis found that gains in business knowledge often fail to translate into sustained earnings growth, particularly among microenterprises, and that observed income effects tend to be short-lived or concentrated in a small subset of firms. The gap between knowing and doing is real, and stubborn.
Subsequent research helps explain why. Interventions that combine training with follow-up support, mentoring, and guidance delivered close to real business decisions, rather than as one-off classroom instruction, tend to show stronger and more durable behavioural effects. The timing and continuity of support matters at least as much as the content.
What matters is not knowledge, but action
This points to a deeper issue, which Michael Frese’s work on Personal Initiative (PI) helps clarify. Drawing on action regulation theory, Frese and colleagues have spent decades investigating why some entrepreneurs act on knowledge while others don’t. Personal Initiative, defined as self-starting, future-oriented, and persistent behaviour, turns out to be a stronger predictor of business success than business knowledge alone.
A landmark randomised controlled trial published in Science found that training focused on personal initiative increased firm profits by 30% over two years among small business owners in Togo, compared to a statistically insignificant 11% for conventional business training. The mechanism matters: it isn’t enough to teach what to do. Entrepreneurs need support in building the habit of acting.
Digital tools falsely assume connectivity
But there is another constraint, and it is structural. Women in low- and middle-income countries remain 14% less likely than men to use mobile internet, according to GSMA’s Mobile Gender Gap Report 2025, and in Sub-Saharan Africa, the gap widens to 29%.
Of the 885 million women still offline globally, around 60% live in South Asia and Sub-Saharan Africa. Affordability, limited digital skills, and safety concerns all constrain adoption. The result? A sector that designs for the connected minority while the majority, women running kiosks, food stalls, and informal retail businesses on a basic handset, are left out of the digital coaching revolution entirely.
Two gaps, one problem
These two gaps, the knowing-doing gap and the connectivity gap, aren’t separate constrains. They’re the same problem, and they require the same solution: coaching that’s continuous, contextual, and accessible on a basic phone call.
This is what we’re testing in Kenya
In partnership with Inkomoko and with support from the Gates Foundation through Shortlist Futures, Fortell Impact is piloting a phone-based AI coaching model for low-income women micro-entrepreneurs in Nairobi, the majority of them urban refugees.
Women interact with a conversational AI agent via a simple phone number, no smartphone, no internet, no literacy-intensive interface required. Rather than delivering information as lessons, the system guides women through real decisions: pricing, cost control, customer follow-up, cash flow.
- It asks reflective questions.
- It prompts next steps.
- It remembers prior conversations.
- It extends the reach of good advisory support, at a fraction of the cost, at any hour, in the user’s own language.
The ambition isn’t to replace human support. Inkomoko’s existing model already combines training with access to finance, market linkages, and ongoing engagement, and that foundation matters. What voice-based AI can do is complement that support, filling the hours and decisions between formal coaching sessions, and making it possible to scale personalised guidance to thousands of women without proportionally scaling costs.
We don’t yet have results.
But the hypothesis is grounded in evidence, both the behavioural science of what drives entrepreneurial action, and the practical reality of what communication infrastructure low-income women in East Africa actually have access to.
If personal initiative training works because it builds the habit of acting, then the question becomes: what’s the most accessible, lowest-friction way to deliver that kind of continuous support at scale? For most of the women we’re trying to reach, the answer is a phone call.
A simple question for the sector
Development organisations spend significant resources designing entrepreneurship programmes for women. The question worth asking isn’t only whether those programmes are effective, but who they’re actually reaching, and whether the tools we use to deliver them are built for the women we say we want to serve.
By Talía Jiménez Romero, Head of Partnerships at Fortell Impact.
