A reading programme runs for a year in a handful of schools. Teachers volunteered for it, the founding team visits every fortnight, and a partner foundation pays for materials. Test scores rise, and the photographs look good. The district office then asks the obvious question: can this go into every school next year?
That question decides whether a promising idea becomes system change or a short-lived success story, and it is often answered too quickly. Ministries, school groups, universities, foundations and EdTech companies across Asia run pilots all the time. Far fewer stop to ask whether the conditions that made a pilot work will still exist once it grows.
This article explains why good pilots often disappoint at scale, what the stronger scale-ups have in common, and how leaders can judge readiness before they commit budgets, teachers and reputations.
Why Good Pilots Often Disappoint at Scale
A pilot is rarely a miniature version of the full programme. It usually runs in unusually favourable conditions. Schools are chosen because their principals are willing. Teachers are often enthusiasts. The people who designed the programme are close at hand to solve problems. Funding per learner tends to be generous, and the fact that someone is watching can lift everyone’s effort.
Remove those conditions and results can shrink or disappear. One of the clearest demonstrations comes from a randomised trial in Kenya, published in the Journal of Public Economics in 2018. Extra teachers on fixed-term contracts raised student test scores when an international NGO hired them. Teachers on identical contracts produced no measurable gain when the government hired them as part of a national roll-out. The researchers traced the gap mainly to implementation delays and to political and bureaucratic opposition, not to differences between the teachers.
The lesson applies well beyond Kenya. The intervention itself did not change. The system delivering it did. Any programme moving from a project team into a ministry, a school network or a franchise model faces the same risk.
Cost is the second trap. The Global Education Evidence Advisory Panel, convened by the World Bank, FCDO, UNICEF and USAID, rates interventions on both learning impact and cost-effectiveness. Analysis of its 2023 report notes that projects tested as small pilots may become much less effective or less cost-effective at regional or national scale. Only a minority of the studies reviewed reported cost data at all. A programme that looks affordable for a few hundred learners may be unaffordable for a few hundred thousand.
What Stronger Scale-Ups Have in Common
The most cited Asian example is Teaching at the Right Level, developed by the Indian organisation Pratham. Children are grouped by their current reading and arithmetic level rather than by grade, and taught with activities matched to that level. J-PAL researchers evaluated variations of the model across India repeatedly over more than a decade. The 2023 GEEAP report later listed it among its “Great Buys”, the most cost-effective approaches it identified.
For school and system leaders, the useful part is how the approach was scaled, more than the approach itself. Pratham and its research partners treated each new setting as something to test, not simply a place to copy the original. Leaders pointed to specific ingredients as essential: grouping by level, explicit learning goals for each group, tailored teaching methods and regular tracking of progress. Other elements were allowed to adapt to government school routines.
The Brookings Center for Universal Education reached a similar conclusion through its Millions Learning research, which followed education initiatives as they scaled in several countries. Practitioners including Pratham and STiR Education worked with Brookings on an education scalability checklist, intended to make scale part of programme design from the start rather than an afterthought.
Pilot Conditions Versus Scale Conditions
Before expanding, compare the conditions under which results were achieved with the conditions that will exist at scale.
| Factor | Typical pilot | Typical scale-up | Question to ask |
|---|---|---|---|
| Schools | Volunteered or hand-picked | Assigned, including reluctant ones | Did the pilot include any ordinary or struggling schools? |
| Teachers | Enthusiasts with extra support | Whole staff, normal workload | Can an average teacher deliver it in normal hours? |
| Support | Founding team on call | District officers or trainers covering many schools | Who solves problems when the designers are not there? |
| Funding | Grant or sponsor | Recurrent budget | What is the cost per learner without the sponsor? |
| Measurement | Research team collects data | Schools report their own data | Can the system track results without outside evaluators? |
| Context | One language, region or school type | Multiple languages, regions and school types | Has it been tested in a second, different setting? |
The Scale Readiness Test: Six Questions Before Expanding
Leaders can use the following six questions as a decision tool. A programme does not need perfect answers to all six, but each weak answer is a risk that should be planned for, not ignored.
- Is the core clearly defined? Write down the few elements that must not change and the elements schools may adapt. If no one can say what the core is, it will be diluted as it spreads.
- Did it work with ordinary implementers? Results achieved only by the founding team or by hand-picked teachers are evidence that the idea can work, not that the system can deliver it.
- Does the cost per learner survive? Recalculate costs without sponsor subsidies, free staff time or donated devices. Include training, replacement, supervision and data collection.
- Can the system measure it without the original team? If only outside researchers can tell whether it is working, problems will go unnoticed once they leave.
- Who owns it after the pilot money ends? A named department, budget line and leader should be responsible before expansion begins.
- Has the effect held in a second, different context? A programme that worked in urban English-medium schools has not yet been tested in rural or multilingual ones.
The same questions apply to education companies. An EdTech product used intensively in a few premium schools is in a pilot phase, whatever its user numbers suggest. Readers interested in that side of the problem may find our earlier analysis of how education companies can prove credibility before they scale useful, along with our guide on how schools and universities should evaluate EdTech evidence.
What to Document at Each Stage
Good scale decisions depend on records kept from the beginning. Many programmes discover too late that they cannot show what they did, where, with whom and at what cost.
- Before the pilot: baseline learning data, the selection method for schools, the intended core elements and the expected cost per learner.
- During the pilot: attendance and participation, how faithfully the programme was delivered, teacher feedback, adaptations made and the reasons for them.
- Before expansion: learning results against the baseline, results for different student groups, actual costs and a written list of what must stay fixed.
- During scale-up: the number of schools, teachers and learners reached, with dates and sources; delivery quality in a sample of sites; and whether learning gains are holding.
This discipline protects institutions in a second way. Scale figures often end up in annual reports, funding proposals and marketing. As our article on how institutions can back up “first”, “largest” and “No. 1” claims argues, a number that cannot be traced to a source is a reputational liability.
Common Mistakes Leaders Make
- Treating enthusiasm as evidence. Positive feedback from pilot teachers is valuable, but it does not show learning improved.
- Scaling the brand, not the method. The programme name spreads to hundreds of schools while the practices that produced results do not.
- Dropping measurement to save money. Monitoring is often the first cost cut at scale, which removes the only way to detect decline.
- Expanding everywhere at once. Phased expansion, with a pause to check results, allows course correction before problems become system-wide.
- Confusing reach with impact. “Used in 500 schools” describes adoption. It does not prove that students learned more.
Reach Is a Milestone; Learning Is the Outcome
That last mistake deserves attention, because reach and impact are both legitimate things to report. They just answer different questions. Schools trained, teachers certified, learners enrolled and districts covered are measurable facts. When those numbers are accurate and well documented, they mark a real institutional milestone. Whether students learned more is a separate question that requires evaluation.
Some scale milestones are exceptional: the largest coordinated teacher-training effort in a region, a learning programme delivered across an unusual number of schools at once, or a curriculum rolled out in a record number of languages. When an achievement like this can be defined precisely, measured consistently and supported by evidence, organisations sometimes seek formal record recognition. Asia Record documents verified record-breaking achievements across Asia. Its assessment checks whether a claim is measurable, standardised and verifiable, and it treats verification as a matter of evidence, not popularity. A school network, university, foundation or EdTech company that becomes an Asia Record holder gains an independent public reference for the milestone itself, listed in the official Asia Record directory of record holders.
The distinction matters. Record certification confirms that a measurable achievement happened as claimed. It does not replace accreditation, regulatory approval or an evaluation of learning impact, and no responsible institution should present it as if it did. The strongest position is to hold both: credible evidence that the programme improves learning, and documented proof of how far it reached.
For education organisations wondering whether a milestone qualifies, the same records described above are what any assessment will ask for: a clear definition, dates, locations, participant counts and supporting documents. Organisations that want to apply for Asia Record can review the requirements and start an Asia Record application. Our step-by-step guide on how to get an Asia Record as a school, university or training provider covers the preparation in more detail. For EdTech firms and private education groups, a verified corporate record of this kind can sit alongside, though never in place of, the learning evidence that buyers and regulators expect.
A Practical Starting Point
Before the next pilot begins, leaders can ask one question that shapes everything else: if this works, what will it take to run it in a hundred schools without us? Designing for that answer from the first day forces clarity about the core method and makes leaders measure costs honestly. It also builds the habit of documentation that later makes both learning results and scale milestones believable.
Across Asia, education systems do not lack promising ideas. What is scarcer is the patience to test whether those ideas survive contact with ordinary schools, tight budgets and many different classrooms. Programmes that pass that test earn something more durable than a successful pilot: the evidence to expand with confidence.