Not everything should scale…
May 27, 2026
By Imelda E. Akurut
Lately, I am convinced that context deserves a budget line, but don’t mind me, I am just a health economist, systems thinker, and reluctant participant in scalability conversations.
Development has become obsessed with scale, but what if not everything should scale?
Today, nearly every major international NGO and donor has a localisation agenda. Reports suggest that over 78% of INGOs now formally commit to locally led development, while global donors increasingly position themselves as champions of “community ownership” and “local voices.” (OECD)
Beautiful language. Inspiring panels. Excellent conference banners. But the money tells a different story; Despite the growing rhetoric, only a small fraction of development funding actually reaches community-based organisations directly. OECD analyses and localisation trackers continue to show that direct funding to local actors remains stubbornly low, often hovering around 10–11%, depending on how “local” is defined.
Still, that is not actually what this article is about; I just wanted to get that out of my chest.
This article is about another development obsession hiding quietly beneath localisation conversations: our addiction to scale.
Because increasingly, even when donors finally decide they want to support locally led initiatives, they come with a condition attached: “Can this be scaled?”
Recently, I came across a funding call from an organisation proudly declaring its commitment to strengthening locally led development. I remember thinking, finally. My people.
Then I opened the application.
Right at the top was the phrase:
“Priority will be given to initiatives with strong potential for scale.”
And there went “my people”
Over the last 15 years, I have interacted with approximately 182 community-based organisations across East Africa. (And yes, before you ask, I really do keep count of the things that interest me.)
Some of the most transformative interventions I have ever encountered would fail a “scalability assessment” within the first ten minutes.
Not because they are ineffective or lack innovation, but because their success is deeply rooted in context, relationships, culture, trust, identity, timing, and people.
Things development frameworks often struggle to measure neatly in Excel sheets.
One organisation I recently interfaced with was tackling teenage pregnancy using music, dance, and drama. Their approach was beautifully local. The youth groups generated income through cultural performances, which then funded community dialogues, outreach activities, and mentorship for young mothers.
In the region where they worked, teenage pregnancy rates hover around 32%, significantly above Uganda’s national average of about 24%.
And the intervention was working.
But here is the important part: its success was not accidental.
The founder is a respected cultural leader in the region. She has deep ties with the Iteso Cultural Union, and expertise is in music, dance, and drama. The community trusts her. Parents listen to her. Young people identify with the cultural format. The intervention fits naturally within the social fabric of that community.
Now comes the favourite donor question:
“Can this work nationally?”
Well.
Can it?
Could you replicate it in another district where cultural structures are different? Maybe.
Could you drop the exact same model into an urban informal settlement in Nairobi or Lagos and expect identical outcomes? Probably not.
Should that make the intervention less valuable? Absolutely not.
And this is where development sometimes confuses replication with effectiveness.
The Project Management Institute defines scaling as the ability to grow while maintaining or improving quality and output. And they also further break down the levels of scale projects can consider ( https://www.pmi.org/disciplined-agile/what-is-scale-and-why-it-is-important).
The next time you are considering scaling, I implore you to ask some questions
· What should actually be scaled?
· Why should it be scaled?
· What value is added by scaling?
Those are important questions.
Questions we often skip because “national rollout” sounds impressive in donor reports.
This is not to downplay the importance of scaling interventions; Sometimes scale works beautifully. Vaccines scale. Bed nets scale. Certain digital systems scale. Medicines scale because biology is wonderfully consistent across geography.
But social interventions? Human behaviour? Culture? Trust? Power? Identity? Those are not software updates.
I learned this lesson personally years ago while working on a digitisation and data management tool in lower-level health facilities. The intervention was doing remarkably well. Data accuracy improved. Reporting delays reduced. Real-time visibility increased. Everyone was excited. Then a funder thought, “Let’s scale this nationally.” Worse still, during the COVID-19 response.
I will spare you the rest of the story, but if you have worked in development long enough, your imagination is probably already accurate.
What worked in a controlled, contextualised environment suddenly collided with infrastructure gaps, varying digital literacy, procurement complications, political urgency, inconsistent electricity, differing workflows, overwhelmed health workers, and the impossible pressure of national expectations.
The intervention did not suddenly become “bad.” It simply stopped being contextually coherent.
And maybe that is the conversation we are avoiding: sometimes scale is not expansion. Sometimes scale is distortion.
To be fair, donors are not irrational for wanting scalable models. Scale promises efficiency. It offers measurable reach. It satisfies boards and taxpayers and annual reports. It creates the seductive possibility of solving large problems quickly.
But development history is filled with examples of large-scale social interventions that became weaker precisely because they moved too far away from the conditions that made them effective in the first place.
Even the OECD now emphasises that effective locally led development requires flexible, context-responsive funding mechanisms rather than one-size-fits-all replication models. (OECD)
So, getting back to where we started, Localization, if we are honest, should mean more than moving money geographically. It should also mean trusting contextual intelligence.
And contextual intelligence is often stubbornly unscalable.
So what am I really saying?
Three things.
1. The inability to scale should not disqualify a good intervention.
A project can be deeply effective, transformative, and worthy of investment without needing to become continental.
Not every initiative must become “the Uber of social impact.”
Sometimes the highest form of success is solving a problem exceptionally well for one community.
2. Localisation without contextual trust is just decentralised control.
If donors truly believe in locally led development, they must resist the urge to standardise every successful idea into a universal model.
Africa is not a monolith.
Neither are districts.
Neither are communities.
What works in Soroti may fail in Kisumu. What succeeds in northern Ghana may collapse in Kigali.
And that is not failure. That is context.
3. We need to stop treating scale as the highest form of legitimacy.
Development has quietly created a hierarchy where:
Local=small, Small = pilot, pilot = temporary, and temporary = less important.
Yet many communities survive because of small, adaptive, trusted local systems that would never survive a donor scalability matrix.
Perhaps instead of asking: “Can this scale?” we should sometimes ask: “Can this remain effective?”
Because in development, there are interventions optimised for scale. And there are interventions optimised for people.
The tragedy is that too often, we force the second category to become the first, and then act surprised when they fail.