Evidence Base
What the field knew in 2022, and what exists now
When the original analysis was published, the ecosystem's knowledge base on scaling ventures was thin, fragmented, and predominantly focused on funding rounds rather than operational dynamics. That characterisation was correct - and it mattered, because the absence of evidence was not neutral. Supply-side solutions proliferated precisely because no one had built the demand-side data infrastructure that would have allowed anyone to test whether those solutions were working.
Four years of subsequent work through the EADC have changed the landscape substantially. State of Startup Innovation reports now exist for Kenya, Ethiopia, and Rwanda. The library that a 2022 interviewee said did not exist now has at least one substantial shelf. The publications produced did not resolve the knowledge deficit; they mapped it with considerably more precision. That is what rigorous inquiry looks like. But it means the frontier has moved."
What that process of building the evidence base taught is that answering questions at scale generates new questions faster than it closes old ones. The publications produced through and alongside the EADC programme did not resolve the knowledge deficit. They mapped it with considerably more precision. That is not failure. It is what rigorous inquiry looks like. But it means the frontier has moved.
What we still do not know - and urgently need to
The 2023–24 capital contraction created the most significant natural experiment the ecosystem has yet produced, and it has not been systematically analysed. The descriptive data - the Series B cliff, the conversion-rate collapse, the M&A wave - tells you the correction was severe and the cliff is real. It does not tell you what distinguished the ventures that survived from those that failed: whether sector, geography, business model, governance quality, capital structure, or founding team composition drove the difference. That data exists, scattered across closure announcements, down rounds, M&A deals, and quietly shuttered operations. It has not been gathered and analysed. Commissioning that retrospective is the most valuable research investment the ecosystem could now make.
The capital structure question is shifting beneath the field even as the descriptive work tries to catch up. Venture debt has moved from marginal complement to structural feature of African startup financing. That shift has implications the current evidence base is not designed to analyse: debt instruments, unlike equity, impose repayment schedules that interact directly with the cash flow patterns of ventures navigating African market conditions. What venture debt actually does to scaling trajectories - whether it extends runway, accelerates professionalisation, or creates repayment pressure that forces premature pivots - is not yet known.
The AI question - whether African ventures participate in AI as builders of African-market products or as consumers of tools built for other markets - has emerged since 2022 with a speed that the evidence base has not kept pace with. The risk is not that African founders fail to adopt AI tools - adoption is accelerating across credit scoring, health diagnostics, logistics optimisation, and SME productivity. The risk is that adoption as consumers of tools built for other markets produces different ecosystem outcomes than participation as builders of AI-enabled products designed for African ones. Understanding the conditions under which African ventures move from AI adoption to AI-enabled product development is the most consequential new research question in the field, and the evidence base to answer it does not yet exist.
The offshore incorporation dynamic has been documented - Offshoring African Startups maps the structural pressures that push founders toward Delaware and the Cayman Islands and demonstrates they are incentive responses rather than founder preferences. What is not yet understood is what specific policy, regulatory, or capital market reforms would meaningfully shift those incentives, and what the ecosystem-level costs of the current offshore default are in terms of tax revenue, talent retention, and local institutional capacity. The diagnosis exists. The counterfactual does not.
The multilateral programming question is harder to ask than to answer. The mechanisms through which heavy donor and multilateral presence reshapes local ecosystems - role collapse, bundled advantage, pricing distortions, crowding out of independent actors - are analytically mapped in When Multilaterals Compete.Practitioner testimony across 60 expert interviews conducted through the EADC is unambiguous on the direction of the effect. What is not yet known is the counterfactual: what ecosystem development trajectories would look like in the absence of heavy multilateral programming. That is the hardest question in the field to answer rigorously, and the one most consequential for the donors and multilaterals who need to hear the answer.
What the donor funding collapse does to ecosystem development trajectories is the most urgent unanswered question of 2026, and the field has almost no analytical tools to address it. The scale of the 2025 contraction places the FCDO's RISA Fund, which primary-funded the EADC programme that produced the evidence base this publication draws on, squarely within the same budget environment the research is trying to analyse.
The geographic gap
The EADC built a rigorous evidence base for Kenya, Ethiopia, and Rwanda. West Africa has no equivalent. Nigeria generates the largest deal count of any single African market, but what explains the gap between deal activity and capital depth at the level of operational firm dynamics - what constraints founders in Lagos, Accra, and Dakar actually face, what distinguishes the ventures that scale from those that do not - cannot be read off the funding data alone. The prescriptions that emerge from East African evidence are being applied across West African contexts by donors, DFIs, and multilateral actors whose programme designs predate the knowledge that would allow them to do this responsibly. That will not stop until the evidence base extends west.
The dissemination problem
Research accumulates. Founders, investors, and ecosystem designers continue to operate on instinct, received wisdom, and anecdote. The mechanism by which evidence reaches and changes the behaviour of the actors with the most leverage over ecosystem outcomes remains almost entirely unsolved. Generating evidence and translating it into practice are different problems that require different institutional designs. The EADC programme invested heavily in the former and incompletely in the latter. The gap between evidence generation and practice change is where the most consequential ecosystem work remains undone - and where this publication tries, imperfectly, to close it.