The Distribution Stack

Wave reached 11 million registered users in Senegal in five years. Ninety percent of the country's adult population. No television campaign did that. No app store feature. Thirteen thousand agents did. They walked Dakar, Thiès, and Saint-Louis neighborhood by neighborhood, signing up the shopkeepers who already sold airtime and cigarettes.

Distribution in most of the world is an attention problem. Ad spend, search ranking, shelf placement. Buy enough impressions and a percentage converts. That math does not work in Sub-Saharan Africa. Discovery here is social. Trust is compound. Scale requires boots on the ground recruiting the people a community already trusts. Three layers. Skip any one and the product stalls.

How people find things

Olga Morawczynski tracked M-PESA adoption in Kibera, one of Nairobi's largest informal settlements, for three years. People learned about the service from family and neighbors, not from advertising. The agents who made it work were not trained financial professionals. They were airtime sellers and kiosk owners already woven into the community's daily routine.¹

The pattern holds at continental scale. Nielsen found that consumers in Africa trust recommendations from people they know above every other channel, and 86 percent of South African consumers prefer discussing products face-to-face over online.² Nanziri and Mukong quantified the mechanism in Uganda: the strongest predictor of mobile money adoption was not income or education but the size of a person's information-exchange network. The more people you talk to regularly, the more likely you are to adopt.³

Daily life in Nairobi, Dakar, or Lagos runs through a structured layer of social interactions. The neighbor who watches your children. The colleague who recommends a mechanic. The cousin who sends airtime as a birthday gift. Every one of these touchpoints is a potential distribution channel. A product that enters through this layer has near-zero acquisition cost after the first referral. A product that bypasses it pays for every user individually, forever.

The trust node

Discovery gets you a name. Trust converts the name into a transaction.

Marcel Fafchamps surveyed manufacturers and traders across Ghana, Zimbabwe, and Kenya and found that fewer than 5 percent used courts or formal legal channels to resolve commercial disputes.⁴ The rest relied on reputation and social pressure. Commerce runs on relational contracts, not written ones. A fabric seller in Abidjan extends credit to a customer she has known for years. No contract. The customer pays because defaulting would cost him his reputation in a market where reputation is the only collateral.

The trust that matters for distribution concentrates in specific people. The pharmacist who triages symptoms because the clinic is too far. The imam whose Friday prayers draw two hundred people. The market woman who has been selling fabric from the same stall for fifteen years. These are the nodes. When they vouch for a product, the social cost of ignoring the recommendation is higher than the effort of trying it.

The first transaction is always small. A 500-CFA mobile money deposit. A single POS withdrawal. The community watches. If the transaction clears, the node's endorsement holds. If it fails, the node's reputation absorbs the damage and no second chance follows. Trust compounds through successful small transactions, not through marketing messages.

Moniepoint understood this in Nigeria. Their 40,000 field agents recruit existing community figures, shop owners and market traders, and pair onboarding with small credit lines underwritten by network reputation rather than credit scores.⁵ The community already trusts the merchant. The product borrows that trust.

Boots and commissions

The trust node does not recruit itself. Someone has to walk into the neighborhood, find the right person, and make a proposition that works.

Wave used dedicated agent openers who moved territory by territory across Senegal. Their job: find the kiosk owner or shopkeeper who already had foot traffic, explain the flat 1 percent fee that undercut Orange Money's tiered rates, and sign them up as a cash-in/cash-out point. The commission structure was the pitch. Every transaction through your kiosk earns you a cut. The trust node becomes a distribution partner because the economics align with the community position they already hold.

An app is cheaper than 40,000 field agents. Wave and Moniepoint deployed the agents anyway, walking neighborhoods before the product had a reputation. Both reached tens of millions of users. Digital-only competitors with bigger marketing budgets are still trying to get past Lagos and Nairobi.

When the network turns

Jack and Suri studied M-PESA's expansion across Kenya and found that agent density was the critical adoption variable.⁶ The relationship was nonlinear. Past a threshold, adoption accelerated without proportional agent growth. Each new agent made the next one cheaper to recruit because the product's presence in the community was already generating referrals.

Wave crossed that threshold in Senegal. Thirteen thousand agents, eight million monthly active users. Once enough kiosks in a neighborhood displayed the yellow logo, new users did not need an agent opener to explain the product. They saw their neighbor using it. Orange Money, which had dominated Senegalese mobile money for a decade, cut its fees by roughly 80 percent. The cut came too late. The density was already compounding.

Moniepoint hit the same inflection in Nigeria. Over one million active POS terminals across all 36 states. Forty-two percent of the country's POS transaction volume.⁷ From 5.2 billion transactions in 2023 to 14 billion in 2025. The growth curve steepened because each terminal in a market or motor park validated the product for the next merchant watching. The same density math that compounds marketplace businesses city by city compounds agent networks neighborhood by neighborhood.

This is where agent banking becomes omni-channel without anyone using the term. The merchant who processes POS transactions also sells airtime, also collects utility payments, also distributes microloans. One trust node, multiple product lines. The same agent mesh that settles daily driver payouts for ride-hail platforms serves as the cash-in/cash-out layer for mobile wallets, insurance premiums, and savings products. Physical presence and digital infrastructure fused through a person the community already trusts.

Skipping layers

Skip discovery and go straight to advertising without the social graph. Acquisition cost never drops because the network does nothing for you.

Skip trust and launch through channels that are not community nodes. Conversion stalls. The app is installed but sits unused. Awareness without credibility.

Skip agents and build digital-only. The pharmacist, the imam, the shop owner never learn the product exists. Without them the social graph has nothing to carry.

The stack follows the same logic as the economy it sits on. Commerce is relational before it is transactional. Trust runs on social graphs, not institutional brands. Scale comes from density, not spend. The businesses that respect this sequence reach 90 percent of a country's adults in five years. The ones that skip layers raise another round to cover acquisition costs that never decline.