The Five Moves

The full framework. Twenty-five cases across the private, public and civil-society sectors.

Move 1 — BUILD

Creating durable institutions, infrastructure and capability rather than advising, advocating or piloting.

Diagnostic question: Does your organization own a piece of infrastructure that others depend on, or does it depend entirely on infrastructure owned by others?

Published September.

Move 2 — COMPOUND

Designing systems where each output becomes the input to the next stage, so effort accumulates instead of resetting.

Diagnostic question: Does each completed cycle of your core activity make the next cycle cheaper, faster or more certain, through a mechanism you can name?

Published October.

Move 3 — INTEROPERATE

Building so that value moves across firms, sectors and borders through shared rails and common standards.

Diagnostic question: When your organization connects to adjacent systems, does it help set the terms of connection, or accept terms set by others?

Published November.

Move 4 — LOCALIZE

Re-deriving capability where you are, rather than renting it from where it was invented.

Diagnostic question: Has your operating model been rebuilt around local constraints, or is it an imported model with local adjustments?

Embrapa
Public sector, state agricultural research corporation
From a 1970s Brazil that imported basic food commodities and whose two million square kilometres of cerrado savanna were considered agriculturally worthless, to the world's leading exporter of soy, beef, maize and cotton, with the cerrado among the most productive farmland on the planet.
Lelapa AI
Private sector, research and product lab
From five African languages spoken by over 300 million people being served poorly or not at all by frontier language models, to InkubaLM: an open, purpose-built small model for isiZulu, Yoruba, Hausa, Swahili and isiXhosa, released in 2024 with the training datasets published alongside it.
Cipla and Indian generic pharmaceuticals
Private sector, enabled by deliberate state IP policy
From triple-therapy HIV treatment costing above 10,000 dollars per patient per year in 2000, pricing most of Africa out of survival, to Cipla's 2001 offer of the same combination at 350 dollars, which collapsed the global price floor and made mass treatment programmes economically possible.
Nollywood
Private and informal sector, organic industry formation
From essentially no domestic film industry in the late 1980s, to one of the largest film industries on earth by volume of output, employing hundreds of thousands, exporting Nigerian culture across the continent and the diaspora, and now supplying original commissions to global streaming platforms.
Debswana and Botswana's diamond bargain
Public-private joint venture
From independence in 1966 as one of the poorest countries in the world with a few kilometres of paved road, to upper-middle-income status, achieved not by discovering diamonds but by repeatedly renegotiating the terms on which they were sold.

LOCALIZE

Move 5 — PERSIST

Operating on multi-decade horizons, with mechanisms designed to protect long-term commitments from short-term pressure.

Diagnostic question: Could your organization continue on its current trajectory if its founder, its principal funder, or the current government were removed tomorrow?

Published December.

Scoring rubric

0 — Absent. The move is not present in any recognisable form.
1 — Espoused. It is stated in strategy or leadership language but nothing operational reflects it.
2 — Mechanized. A named mechanism exists, is resourced, and runs without leadership attention.
3 — Compounding. The mechanism strengthens itself: each cycle makes the next one cheaper or more certain.

0–5: Project mode. The organization is executing discrete initiatives. Effort does not accumulate into position, and performance depends on continued leadership attention. This is the ordinary state of most organizations in the study before their trajectory changed.
6–9: Transitional. At least one move is mechanized and beginning to hold without attention. The characteristic risk at this stage is stalling: mechanisms exist but the deposits leak, so the curve flattens rather than compounds.
10–15: Institutional trajectory. Multiple moves are mechanized and at least one compounds. The organization is likely to survive a leadership or funding transition. The binding question is no longer capability but persistence.

Entry sequences

New organization, thin resources

1. LOCALIZE — With no balance sheet to defend, the only cheap advantage available is a model designed against a constraint incumbents have adapted around rather than solved.
2. COMPOUND — Thin resources make cycle speed the decisive variable; the cases show small organizations overtaking larger ones by running the loop more often, not by running it better.
3. BUILD — Build only once the localized model has proven demand, and build the narrow layer that model depends on — the sequence Twiga and M-KOPA both followed.

Existing trust network or member base

1. COMPOUND — The membership already generates a record; the first task is to capture it institutionally rather than let it dissipate, as KTDA did with grower-level data.
2. BUILD — Trust converts into position only when it is attached to an asset members must route through, which is what turned tea growers into factory owners.
3. PERSIST — Member-based organizations fail at succession more than at strategy; institutionalize governance before the founding cohort turns over.

Regulator or state actor

1. INTEROPERATE — Rule-setting authority is worth most when it is exercised on connection terms, as in Nigeria's switching standards and the AfCFTA settlement layer.
2. BUILD — A shared public layer with a usage mandate — the Irembo pattern — converts regulatory authority into permanent infrastructure.
3. PERSIST — State initiatives die with administrations; Botswana's fiscal rule shows that statutory basis, not enthusiasm, is what carries a reform across cycles.

Resource or sector incumbent

1. BUILD — Incumbents have the one thing building requires and startups lack: the ability to absorb a cost that only pays back once the layer is load-bearing.
2. INTEROPERATE — The built layer is only worth its cost if others connect to it; Interswitch's neutrality shows what terms make that possible.
3. COMPOUND — Incumbents leak their operational record to vendors and partners more than any other group; closing that leak is usually the highest-return move available.

Coalition or alliance builder

1. INTEROPERATE — A coalition's only asset is the terms of connection between members; define them narrowly and operationally, as the Aadhaar interface did.
2. PERSIST — Coalitions dissolve when the convening funder rotates; a fixed cadence and independent governance, as with the Ibrahim Index, is what makes them durable.
3. COMPOUND — Each member interaction should deposit something the coalition owns — a standard, a dataset, a method — or the coalition remains a meeting.

The 90-day entry protocol

Days 1 to 15 — Locate the constraint
Trace ten real failed transactions or stalled decisions end to end. Record the first point of breakage in each. Do not commission research; do the tracing with the people who own the work.
Days 16 to 45 — Name one mechanism
Select the single lowest-scoring move from your profile and define one mechanism that would move it from espoused to mechanized. It must have an owner, a budget line and a cycle time measured in weeks.
Days 46 to 75 — Run the loop three times
Execute the mechanism at least three times. Three cycles is the minimum needed to distinguish a working mechanism from a successful first attempt driven by leadership attention.
Days 76 to 90 — Fix ownership of the deposit
Establish where the residue of each cycle accumulates and who holds title to it. Move the record, the method or the relationship in-house before the mechanism scales.