How Governments Should Buy Acceleration

Governments should buy acceleration the way they buy roads: through competitive tender to qualified independent operators, with payment tied substantially to verified outcomes — venture survival, revenue growth, and jobs at 24 months — rather than to activities delivered. The evidence from World Bank, OECD, and a decade of payment-by-results experience is consistent: states get better results contracting demanding outcomes from third parties than running programs themselves (1, 2, 8, 9). Yet most public procurement of entrepreneur support in East Africa still pays for cohorts run and founders trained, which is why so much of it buys workshops instead of businesses.

Key Takeaways

• With donor funding receding after the 2025 aid collapse, government is now the largest remaining buyer of entrepreneur support in Uganda, Kenya, Rwanda, and Tanzania — how it buys will shape the region’s ESO industry for a decade (3, 11).
• The evidence base is double-edged: GALI data on 23,000+ ventures shows acceleration works on average, but 2025 NBER analysis finds most individual programs add negative value — meaning an undiscriminating public buyer is statistically likely to purchase a value-destroying program (4, 5).
• World Bank/infoDev and OECD reviews both conclude governments achieve more by competitively funding independent operators against defined outcomes than by operating incubation programs directly (1, 2).
• A decade of payment-by-results evidence warns against naive outcome contracting: verification costs are real, gaming is real, and pure PbR performs no better than well-managed conventional contracts — the design answer is hybrid payment, independent measurement, and realistic timelines (8, 9).
• Uganda already hosts working reference models: the JICA-MTIC NINJA growth-stage accelerator selected 10 ventures from over 130 applicants against hard revenue criteria, and the NSSF Hi-Innovator program has backed 438 businesses while generating 14,000 new pension contributors — a buyer purchasing its own future customers (6, 7, 10).
• The Demanding Customer Contract — five clauses covering outcomes, competition, verification, time, and capability transfer — gives ministries and donors a usable template for the next tender they issue.

Why Are Governments Now the Largest Buyer of Acceleration?

For twenty years, the question of how to procure entrepreneur support barely mattered to East African governments, because somebody else was paying. Bilateral donors and foundations funded the accelerators, the hubs, and the training programs; ministries cut ribbons. That world ended in 2025. The dismantlement of USAID and the broader Western aid retreat removed the sector’s largest funder in a matter of months, and the consequences for entrepreneur support organizations were severe — a sorting event I have examined in detail in the ESO funding crisis in Africa (3).

What remains on the demand side is the state, in three forms. First, direct national programs: Uganda alone hosts the JICA-partnered NINJA Acceleration Program run with the Ministry of Trade, Industry and Cooperatives, the NSSF Hi-Innovator scheme, and an emerging National Startup Policy developed with ecosystem bodies like Startup Uganda (6, 7, 11). Second, development-bank and bilateral money that increasingly flows through government priorities rather than around them. Third, public pension and sovereign capital — NSSF’s progression from grant-maker to fund-of-funds architect being the regional landmark (10).

This concentration of buying power is not a misfortune. It is an opportunity, for one reason: a single demanding customer can reform a market faster than a hundred polite ones. When the dominant buyer of acceleration starts paying for verified outcomes, every operator in the market re-engineers around outcomes within two procurement cycles. When the dominant buyer pays for attendance sheets, the market produces attendance sheets. East Africa’s ESO industry will become whatever its largest customer rewards. That is the entire stake of this argument.

What Goes Wrong When Governments Buy Activities Instead of Outcomes?

Walk through a typical public tender for enterprise support anywhere in the region and you will find the same architecture: a budget, a target number of “beneficiaries trained,” a list of mandated workshop topics, a delivery period of twelve to eighteen months, and reporting requirements built around disbursement and headcount. Nothing in the contract pays more if the supported ventures survive, grow, or hire — and nothing pays less if every one of them is dead within a year.

This design fails in four predictable ways.

It cannot distinguish good operators from busy ones. The global evidence makes this failure expensive. GALI’s decade of data across 23,000+ ventures shows accelerated companies outperform rejected peers on revenue, employment, and capital raised (4). But the 2025 NBER working paper “Beyond Demo Day” finds that most accelerators have negative value-added against a no-accelerator benchmark — the average gains come from a small right tail of excellent programs (5). I have unpacked that distribution in whether startup acceleration works at all; the procurement implication is brutal. A government that buys acceleration without outcome evidence is drawing randomly from a distribution whose median product destroys value. Activity-based contracts make that blindness permanent, because they never generate the data that would reveal which tail the operator occupies.

It selects for proposal-writing, not program quality. When payment follows activities, the competitive weapon is the bid document. The organizations that win are those with the best grant writers and the most compliant logframes — capabilities that correlate with nothing GALI measures. Operators who invest in mentor depth, capital connection, and post-program support are spending money the contract does not reward.

It compresses time horizons below the production cycle of the product. Venture outcomes take 24 to 36 months to materialize; most public contracts run 12 to 18. The result is a sector trained to optimize what can be photographed before the contract closes — demo days, certificates, launch events — rather than what matters after it. The deepest damage of activity procurement is temporal: it teaches a whole industry that its product is the cohort, when its product is what the cohort produces two years later.

It builds no national capability. Activity contracts treat each program as a disposable project. When the contract ends, the curriculum leaves with the international contractor, the data is archived in a donor report, and the next tender starts from zero. After two decades of enterprise-support spending in East Africa, remarkably little institutional capability has accumulated in-country — not because the money was small, but because nothing in the contracts required it to stay.

The accountability dimension of this failure — what citizens and parliaments should demand from public startup programs — deserves its own treatment, which I give it in government accelerator programs and accountability. Here the focus is the buyer’s side of the table: how to write the contract right in the first place.

What Does the Evidence Say About Outcome-Based Contracting?

The instinctive correction — “stop paying for activities, pay for results” — is correct in direction and dangerous in naive form, and an honest manifesto has to hold both truths.

The direction is well supported. World Bank/infoDev’s foundational review of business incubation policy concluded that governments do better funding independent, professionally run operators than operating programs through public agencies, whose incentives, salaries, and political cycles fit badly with venture timelines (1). The OECD’s 2024 review of incubation and acceleration policy points the same way: public funders achieve more as structurers and evaluators of third-party programs than as deliverers (2). And the structural logic is hard to argue with: acceleration is a service with measurable outputs, delivered competitively by specialized firms — exactly the category of spending where outcome-linked, competitively tendered contracts outperform in-house provision.

The caution comes from the payment-by-results decade. DFID made PbR a flagship strategy in 2014, “sharpening incentives to perform” across health, education, and economic development programming (8). The accumulated evaluation evidence, synthesized a decade later, is humbling: no systematic evidence that PbR produces more innovation or better results than well-managed conventional contracts; real verification costs; documented gaming, including providers chasing easy-to-reach beneficiaries to hit payment triggers; and risk premiums that smaller local providers cannot carry, which perversely concentrates contracts among large international firms (9). The Girls’ Education Challenge evaluations found exactly the distortion an accelerator buyer should fear: pressure to maximize countable beneficiaries crowding out depth with the hardest cases (9).

The synthesis is not to abandon outcome contracting but to design it like an adult: hybrid payment rather than pure contingency, independently verified metrics rather than self-reported ones, time horizons that match the product, and deliberate protection for capable local operators. That is what the framework below does.

The Demanding Customer Contract: Five Clauses for Buying Acceleration Well

Every acceleration tender a ministry, development agency, or public fund issues should be testable against five clauses. Together they form what I call the Demanding Customer Contract — the procurement standard a government adopts when it decides to be a customer of its ESO industry rather than a patron of it.

Clause 1 — The Outcomes Clause: pay a meaningful share against verified venture outcomes at 24 months. Not 100%: the PbR evidence says pure contingency pricing pushes risk onto operators, inflates bids, and excludes local firms (9). The workable structure is roughly half fixed (mobilization and delivery), with the balance staged against pre-agreed outcome triggers — two-year venture survival, audited revenue growth against intake baseline, employment created, follow-on capital or contracts secured. The triggers must be few (three to five), defined at signature, and benchmarked against the venture profile actually being served. A growth-stage program like Uganda’s NINJA — which required applicants to show roughly UGX 500 million in annual revenue and selected 10 ventures from over 130 applicants — should carry harder revenue triggers than a first-time-founder program in a secondary city (6). Outcome pricing without intake honesty is just risk transfer with extra steps.

Clause 2 — The Competition Clause: tender competitively to independent operators; never operate in-house. The government’s job is to be a demanding customer, not a program director. Open tenders to any operator — local or international — that can produce evidence of past venture outcomes, and weight that evidence above proposal aesthetics. Pre-qualification should ask one question above all: show us the two-year survival and revenue data for your last three cohorts. Operators who cannot answer are not penalized for honesty — they are offered smaller, first-rung contracts where they can build the record. This single question, asked consistently by the region’s largest buyer, would do more to professionalize East African ESO measurement than a decade of donor toolkits.

Clause 3 — The Verification Clause: independent measurement, intake baselines, and public reporting. Self-reported alumni surveys are marketing. The contract should fund — separately, at 3–5% of contract value — an independent verifier that captures baseline data on every admitted venture and, critically, on a comparison pool of qualified non-admitted applicants, GALI-style (4). This is what converts a procurement exercise into a learning system: after three contracted cohorts, the buyer knows not just whether the operator hit triggers, but whether the program beats its counterfactual. All outcome data should be published. Public money, public numbers.

Clause 4 — The Time Clause: contract in 3–5 year frames with break points, not 12-month projects. If outcomes mature at 24 months, a 12-month contract is structurally incapable of paying for them. The honest structure is a multi-cohort framework contract — three to five years — with annual break points tied to delivery quality and a final settlement tied to verified outcomes. This also solves the operator’s side of the problem: no serious institution can hire mentors, build alumni infrastructure, or carry post-program support — the phase where, as I have argued in the post-accelerator valley of death, East African ventures actually live or die — on a sequence of one-year stop-start projects.

Clause 5 — The Transfer Clause: every contract must leave capability in the market. Curriculum, measurement systems, and venture data developed under public contract should be licensed to the buyer and remain in-country. International operators should be required to bid with local delivery partners on a defined capability-transfer schedule — not as decoration, but with payment milestones attached to the local partner independently delivering program components by year three. The endgame of public acceleration procurement is a domestic industry of operators good enough to win the contracts outright. A buyer who never builds its supplier base will be importing acceleration forever.

What Should Government Buyers and Operators Do Now?

For the ministry or agency official, the next tender is the reform. Insert the pre-qualification question from Clause 2. Restructure payment 50/50 fixed-to-outcomes with three triggers you can verify. Budget the independent verifier. Extend the frame to 36 months even if the first cohort is the only committed spend. And resist the procurement office’s instinct to maximize beneficiary counts — five hundred founders trained is not a result; fifty businesses alive and growing at month 24 is.

The region’s own reference models show the direction is practical, not theoretical. The NINJA program demonstrates selective, criteria-driven intake under a bilateral-ministerial structure (6). Hi-Innovator demonstrates something even more important: a closed incentive loop, in which Uganda’s pension fund backs small businesses and harvests the outcome it actually wants — 438 businesses supported, and 14,000 new contributing pension members generated by the formalization and hiring that followed (7, 10). When the buyer’s own balance sheet benefits from venture success, outcome orientation stops being a reporting requirement and becomes self-interest. That is the deepest version of demanding-customer procurement available, and East Africa invented it at home.

For operators, the manifesto reads in reverse, and it is bracing. If governments become demanding customers, the operators who thrive will be those who can already answer the survival-and-revenue question — which means building the outcomes file now, on your own initiative, before any tender requires it. Track every graduating venture for 24 months. Capture intake baselines. Reconstruct historical cohorts where you can. Then bid with the confidence of a firm selling a measured product rather than a proposal. Operators who fear outcome contracting are telling you something about their product; operators who lobby for it are telling you something better.

A final word to donors, who have not left the stage entirely: the highest-value use of remaining grant capital is not running parallel programs but underwriting this market’s formation — funding the independent verification infrastructure, capitalizing local operators so they can carry outcome risk, and co-financing government tenders that adopt these clauses, as ANDE’s donor-practice work urges (12). Aid that builds a procurement market dies well.

What Would Success Look Like by 2030?

A functioning East African market for acceleration, purchased rather than donated. Concretely: every major public enterprise-support tender in Uganda, Kenya, Rwanda, and Tanzania carrying outcome triggers and independent verification; a published regional dataset of program-level survival and revenue results, comparable across operators; framework contracts long enough that operators invest in post-program support; at least a dozen local operators with three-cohort outcome records strong enough to win against international bidders — and the weakest third of today’s providers exited, not by scandal, but by the quiet arithmetic of unrenewed contracts.

None of this requires new money. It requires the largest buyer in the market to start behaving like a customer: specifying the product, verifying delivery, paying for results, and developing suppliers. Governments that buy roads this way get roads. Governments that buy acceleration this way will get what the region has needed all along — not more programs, but more companies. That is a procurement reform worth being excited about, and it can start with the next tender someone reading this is about to sign.

Frequently Asked Questions

Should governments run their own accelerators?
No. World Bank/infoDev and OECD reviews consistently find governments achieve better results competitively contracting independent operators against defined outcomes than operating programs themselves. Public agencies’ pay scales, political cycles, and risk incentives fit venture support badly. The state’s highest-value role is demanding customer: specifying outcomes, verifying results, and paying accordingly (1, 2).

What outcomes should an acceleration contract pay for?
Three to five verifiable triggers fixed at signature: venture survival at 24 months, revenue growth against intake baseline, employment created, and follow-on capital or contracts secured. Triggers must match the venture profile being served — growth-stage programs carry harder revenue tests than first-time-founder programs — and be independently verified, never self-reported (4, 6).

Doesn’t payment-by-results just shift risk onto operators?
It can, which is why pure contingency contracts fail. A decade of PbR evidence shows verification costs, gaming, and risk premiums that exclude smaller local providers. The workable design is hybrid: roughly half fixed payment for delivery, the balance staged against outcome triggers, with independent verification funded separately (8, 9).

Why do contracts need to run three years or longer?
Because venture outcomes mature at 24–36 months, and a 12-month contract structurally cannot pay for them. Short projects train operators to optimize what is visible before closeout — events and certificates — rather than survival and growth afterward. Multi-cohort framework contracts with annual break points align contract time with product time (5, 9).

What is the Demanding Customer Contract?
A five-clause procurement standard for public buyers of acceleration: pay substantially against verified 24-month outcomes; tender competitively to independent operators; fund independent measurement with intake baselines and public reporting; contract in 3–5 year frames; and require capability transfer so curriculum, data, and delivery skill remain in the local market.

Related Reading

• The ESO funding crisis in Africa: after the aid collapse
• Does startup acceleration work? What the GALI and NBER evidence actually says
• Government accelerator programs and the accountability question
• The post-accelerator valley of death in East Africa

Sources and Evidence

1. World Bank / infoDev. “Business Incubation: Lessons for Policymakers.” https://documents1.worldbank.org/curated/en/981161468331855750/pdf/700230ESW0P11100Business0Incubation.pdf — Foundational multilateral evidence that public funding works best through competitively selected third-party operators.
2. OECD, 2024. “Start-up globalisation through incubation and acceleration.” https://www.oecd.org/content/dam/oecd/en/about/projects/cfe/incubation-and-acceleration/Start-up-globalisation-through-incubation-and-acceleration.pdf — OECD policy review supporting outcome-oriented contracting over state operation of programs.
3. Oxfam America, 2025. “What did USAID do and what are the effects of USAID cuts?” https://www.oxfamamerica.org/explore/issues/making-foreign-aid-work/what-do-trumps-proposed-foreign-aid-cuts-mean/ — Documentation of the ~90% contract termination that made government the sector’s largest remaining buyer.
4. Global Accelerator Learning Initiative (GALI / ANDE / Emory University). “Does Acceleration Work?” https://andeglobal.org/publication/does-acceleration-work/ — The largest longitudinal acceleration dataset (23,000+ ventures), with comparison-group methodology any public verifier can adopt.
5. National Bureau of Economic Research, 2025. “Beyond Demo Day” (Working Paper 35063). https://www.nber.org/papers/w35063 — Academic analysis finding negative median accelerator value-added with a high-performing right tail; the core argument for evidence-gated procurement.
6. Hindsight Ventures. “JICA NINJA Acceleration Program — Growth Stage, Uganda.” https://hindsightventures.co/flagship-programs/jica-ninja-acceleration-program — Implementer’s program record: 10 ventures selected from 130+ applications against hard revenue criteria (~UGX 500M annual revenue threshold).
7. Ministry of Trade, Industry and Cooperatives (Uganda). “Launch of JICA NINJA Acceleration Program for Growth Stage in Uganda.” https://www.mtic.go.ug/launch-of-jica-ninja-acceleration-program-for-growth-stage-in-uganda/ — Official government source on the bilateral program structure.
8. UK Department for International Development, 2014. “Payment by Results Strategy: Sharpening incentives to perform.” https://www.gov.uk/government/publications/dfids-strategy-for-payment-by-results-sharpening-incentives-to-perform/payment-by-results-strategy-sharpening-incentives-to-perform — Primary source for the flagship donor PbR strategy.
9. Clist, P., 2019. “Payment by results in international development: Evidence from the first decade.” Development Policy Review. https://onlinelibrary.wiley.com/doi/abs/10.1111/dpr.12405 — Peer-reviewed synthesis of PbR evidence: limited performance advantage, verification costs, and gaming risks that hybrid contract design must answer.
10. ImpactAlpha, 2025. “Pension fund in Uganda readies a $100 million fund-of-funds to create jobs — and savers.” https://impactalpha.com/pension-fund-in-uganda-readies-a-100-million-fund-of-funds-to-create-jobs-and-savers/ — Source for Hi-Innovator’s 438 businesses backed and 14,000 new pension contributors; the closed-loop buyer model.
11. UNCDF. “Doing more together: the coming of age of Startup Uganda as an ecosystem builder.” https://www.uncdf.org/article/8206/doing-more-together-the-coming-of-age-of-startup-uganda-as-an-ecosystem-builder — UN agency documentation of Uganda’s emerging National Startup Policy process.
12. ANDE, 2024. “Sustain Impact: Donor practices to grow enterprise support organizations.” https://andeglobal.org/publication/sustain-impact-donor-practices-to-grow-enterprise-support-organizations/ — Sector-body guidance on donors funding market formation and institutions rather than perpetual project cycles.