Most scale-up plans have a column called “responsible” and one word in it: Ministry.
That is the first sign the plan will stall. A ministry is not a person, and it is not even a single organisation. It is a planning directorate, a finance office, a human resources department, a procurement unit and a set of regional offices, each with its own budget, its own deadlines and its own reasons to say no. When a plan assigns a step to “the Ministry,” it assigns it to nobody in particular.
We have written before about why universal health coverage pilots stall after the pilot phase. This piece is the practical counterpart. The thesis: scaling a health pilot into a national program is mostly a sequence of handovers to institutions outside the health program itself, and each handover needs a named owner who has both the authority and the incentive to take it. The clinical model is usually the easy part.
Why pilots are built to succeed and not to scale
Start with the incentives, because they explain most of what follows.
A pilot is judged on whether it works. Its funder wants a clean evaluation, its implementing partner wants a publishable result, and its staff want the model delivered as designed. Every one of those incentives pushes toward protecting the pilot from the ordinary state: separate money, separate hiring, separate supplies, separate data, and supervision by people who care deeply about the outcome.
Those protections are exactly what make the result credible. They are also exactly what must be dismantled to scale it. The pilot proves that a model works when insulated from the system. Scale asks whether it works when the system runs it. Those are different questions, and the evidence from the first rarely answers the second.
Nobody inside a pilot is paid to prepare for that second question. The evaluation ends, the grant closes, the implementing partner moves to the next project. The handovers that decide whether the program survives fall into the gap between a donor’s funding cycle and a government’s budget cycle. That gap is where most good pilots go to die.
A reference point: Ethiopia’s health extension workers
It helps to look at a program that did scale. In 2003, Ethiopia’s Ministry of Health established the Health Extension Program to bring basic preventive and curative services to rural communities. The design was simple to describe: each kebele, the lowest administrative unit, typically serving a few thousand people, would get a health post staffed by two health extension workers trained for about a year. According to published accounts of the program, including from JSI’s L10K project, more than 33,000 of these workers were trained and deployed nationally.
The program has well-documented weaknesses, including uneven quality between regions and heavy workloads. But the scale itself is instructive. It did not happen because a pilot was replicated district by district. It happened because the program was built into the machinery of the state: a recognised cadre, a training pipeline, a place in the administrative structure down to the smallest unit. A 2019 review of the program’s first fifteen years, published in Globalization and Health by Yibeltal Assefa and colleagues, describes the workers as salaried government employees, assigned to health posts after their training. That detail matters more than it looks. It means the treasury, not a donor, carries the payroll.
That is the core of the playbook.
Step 1: Turn the budget from project to line item
Owner: the finance ministry’s budget office, with the health ministry’s finance director.
A pilot is funded by a grant. A national program needs a line in the recurrent budget that survives a change of government. The health ministry cannot create that alone. The budget office must accept that the cost is permanent, and it will only do so if the costing is honest: the cost of reaching the remote and hard-to-serve, not the pilot average.
The practical move is to cost the program at full scale before the pilot ends, and to start moving a share of recurrent costs onto the government budget during the pilot itself. A program that has never been paid for by the treasury is still a donor project, however successful.
A back-of-envelope costing
The arithmetic of that honest costing is worth doing, because it is where many plans quietly break. The figures below are illustrative assumptions, not data from any country.
Suppose a community health pilot covers 20 districts at a cost of $4 per person per year, and the country has 200 districts. Multiply by ten and the scale-up looks like a tidy number. It is not.
Pilot districts are usually chosen because they are accessible, have functioning facilities and a cooperative local administration. The remaining districts include the mountains, the islands and the places with no road in the rainy season. Transport, supervision and staff retention cost more there. If those districts cost 50 percent more to serve, and they make up half the national population, the national average rises from $4 to $5 per person. That is a 25 percent gap in the budget request, before anyone has argued about salaries.
Then add what the pilot got for free. Donor-funded vehicles that will need replacing. Technical assistance staff who did the planning. The evaluation team whose data collection doubled as supervision. Put those in and the true figure can climb further. A finance ministry shown the $4 figure and later asked for more will stop trusting the health ministry’s numbers. Shown the higher figure first, it may still say yes, and mean it.
Step 2: Create the job, not just the training
Owner: the civil service or public service commission.
Many pilots rely on staff hired on contract through a project. At scale, those people need real posts: a job title, a salary grade, a career path and pension rights. Only the body that governs public employment can create them. This step is slow, political and often the single biggest delay, because every new cadre raises questions from existing ones about pay and status.
Ethiopia’s program worked partly because health extension workers were built into the public workforce rather than left as volunteers or project hires. That choice has a cost: a salaried cadre is a permanent wage bill, and it creates expectations of promotion and pay rises that a volunteer model avoids. But it also creates the thing that volunteer models struggle to sustain, which is a worker who is still in post five years later.
Step 3: Move procurement into the national system
Owner: the central medical stores or national procurement agency.
Pilots often buy their supplies separately, with donor procurement rules and a dedicated logistics contractor. At scale, the program’s commodities have to travel through the ordinary national supply chain. That chain needs to know the program’s products, forecast them and include them in its tenders. If it cannot, the program will stock out the moment the parallel system is switched off.
The trap here is timing. Central stores usually forecast a year or more ahead. A program that asks to be included only when the donor contract ends will face a gap of months with nothing on the shelves.
Step 4: Put the data into the routine system
Owner: the health information unit.
A pilot’s data usually come from a special survey or a custom database, run by the evaluation team. A national program has to report through the routine health information system that every facility already uses. That means agreeing a small number of indicators, adding them to national forms or software, and accepting that routine data will be messier than the pilot’s.
Step 5: Hand supervision to the regions
Owner: regional or provincial health authorities.
In a pilot, central staff or consultants supervise. At national scale, that is impossible. Regional health offices need the budget, vehicles and staff to supervise the program themselves, and it needs to be part of how they are assessed. If supervision stays central, it thins out and eventually stops.
Step 6: Keep a small central team with teeth
Owner: a named program director in the health ministry.
After the handovers, someone must still own the whole. That person needs authority to escalate problems across institutions and a mandate that survives a change of minister. This is where the “Ministry” in the plan should become a name.
The strongest counterargument
A fair critic would say this playbook is too tidy. In many countries, the handovers are not blocked by a missing name in a plan. They are blocked by a fiscal crisis, a hiring freeze imposed by a lender, a civil war or a political leadership that simply does not prioritise primary care. No amount of named ownership will make a treasury fund a program it cannot afford.
The critic is partly right. Named owners are necessary, not sufficient. Where fiscal space is genuinely absent, the honest answer may be a slower scale-up, a narrower package, or a longer transition in which donors pay a declining share on a published schedule.
But the counterargument is often used to excuse the failures that are within reach. Plenty of programs stall in countries with adequate budgets, because nobody asked the civil service commission for posts until the grant had three months left. Fiscal constraints set the ceiling. Unclear ownership is usually why programs fall well below it.
Second-order effects of a successful handover
Success creates its own problems, and a good plan anticipates them.
A new salaried cadre changes the labour market for health workers. Existing cadres, such as nurses or clinical officers, may press for pay parity, and unions may object to task-sharing. That pressure can raise the wage bill well beyond the program’s original costing.
Absorbing a vertical program into the routine system also dilutes attention. The health information unit now carries a few more indicators among hundreds. The central stores handle a few more products among thousands. That is the price of permanence: the program is no longer special, which is exactly why it survives, and also why its quality may drift.
Finally, a program that becomes part of the state becomes part of its politics. Posts get allocated with an eye to constituencies. Budgets get cut in lean years. A program director with a mandate that outlasts a minister is the main defence.
What each actor should do differently
Health ministries should write the scale-up plan with a named person and institution against each handover before the pilot starts, and review it every quarter.
Finance ministries should ask for full-scale, hard-to-reach costing up front, and agree a transition schedule in which the government share rises year by year during the pilot.
Donors and funders should budget for the handover itself: the civil service negotiations, the procurement forecasting, the data integration. These are unglamorous, cheap relative to the program, and rarely funded. A grant that ends without them has paid for a result nobody can keep.
Implementing partners and founders running pilots should design for the version of the model that ordinary institutions can run, and test that simpler version before claiming the pilot is ready.
Clinicians and district managers should be asked, before scale, which parts of the model they could not sustain without extra help. They usually know.
The uncomfortable trade-off
Every handover costs fidelity. The treasury will trim the budget. The civil service will set salaries lower than the project did. Routine data will be less precise. Regional supervision will be patchier than the consultants’ visits.
Programs that insist on pilot-level fidelity rarely scale. Programs that scale accept a simpler, sturdier version of the model that ordinary institutions can run. Deciding in advance which parts of the model are non-negotiable, and which can be simplified, is the most important design decision a scale-up team makes.
What to watch over the next two years
With external health financing under pressure in many countries, more programs will face transitions sooner than planned. Watch which governments publish transition schedules with named owners, rather than hoping the next grant arrives. Watch whether community health workers in donor-backed programs are moved onto public payrolls or left as volunteers, since that single choice predicts much of what follows. And watch the procurement lists of central medical stores: when a program’s products appear in national tenders, the handover is real.
The takeaway
Scale is not a bigger pilot. It is a series of handovers to finance, the civil service, procurement, information systems and regional authorities. Write the plan with a person and an institution against each one, and start the handovers before the pilot ends. If the only name in the plan is “the Ministry,” the plan is still a hope.
Which handover was hardest in a program you helped scale: money, staff, supplies or data? We’d like to hear what got it unstuck.
