The float is the product
I spent four years building on Kenya's mobile-money rails before I understood what I was actually building on: not a payments company, but something much closer to a narrow bank.
The first time I saw the float number, I assumed it was a typo. I was three weeks into a job wiring a lending product into a wallet API, and someone in treasury pulled up a dashboard showing the aggregate balance sitting in customer wallets overnight. It was larger than the deposit base of several licensed banks in the same market. Nobody in the room thought this was remarkable. To them it was plumbing.
That number is the thing I keep coming back to. We talk about mobile money as a transfer mechanism — send money home, pay the shop, top up airtime. But the economically interesting object is not the transfer. It is the balance that never moves: the shillings parked in a hundred million wallets because moving them out costs something and leaving them in costs nothing.
Regulators call this the trust account. Operators call it float. Whatever the name, it is a deposit book that arrived without a banking licence, and the last three years of rulemaking across East Africa are best read as an attempt to decide who owns the interest on it.
A deposit book that arrived without a banking licence — and the rulemaking is really about who owns the interest on it.
There is a second-order effect that I think is underrated. Because float is sticky and cheap, it subsidises everything built on top of it: agent commissions, free P2P tiers, the loss-leading merchant rates that pushed cards out of the small-ticket market. Take the float yield away — pass it to customers, as one draft rule proposes — and a lot of the consumer-facing economics stop working at current prices.
What this means if you are building
If your product's unit economics assume today's wallet pricing, you are underwriting a regulatory position, not a market one. The teams I've seen handle this well did two unglamorous things: they modelled a world where P2P is priced at cost, and they built the ability to hold customer balances themselves, under whatever licence their market offers.
Neither is a strategy. Both are options. In a market where the rules are being written in real time, options are most of what you can buy.
- Central Bank of Kenya, monthly mobile payments statistics — placeholder for the real series.
- Operator annual reports, FY2025 segment disclosures.
- Figures in this piece are indicative and should be replaced before publication.
One chart, one argument, most Fridays.