Vodafone built a tool for microloan repayment, watched pilot users repurpose it to trade money for money, and shipped what they were actually doing instead

the problem

A pilot product built for one specific use often reveals a completely different, larger unmet need through how users actually misuse or repurpose it

background

what everyone would do

Correct the off-label behavior and steer the pilot back to its funded purpose, or scrap the narrow tool entirely and build a full banking product from scratch — the two responses most teams give to a pilot that isn't being used as designed. Both discard the same information: users had already shown, unprompted, exactly what they wanted, using whatever imperfect materials were at hand.

what they saw

Users reselling airtime as an improvised currency weren't misusing the tool, they were revealing a need more precisely than any market research could — a demand for sending value directly to another person, not for repaying a loan. That they tolerated a lossy, inconvenient workaround to get it was the strongest possible evidence the underlying demand was real and large, not a reason to correct their behavior back to the intended use case.

the move

Vodafone and Safaricom rebuilt the pilot's original narrow tool into M-Pesa, a direct phone-to-phone cash transfer and storage system, formalizing the exact peer-to-peer transfer behavior users had already improvised rather than continuing to push the original microloan-repayment framing. M-Pesa launched commercially in Kenya on 6 March 2007.

why it works

Pilot users without a direct way to send money used the closest available substitute — transferable phone airtime — despite the loss taken reselling it for cash, which functioned as a revealed preference far more convincing than a survey answer. Building the actual capability directly, phone-to-phone cash transfer without the airtime detour, removed exactly the friction users had already shown they'd tolerate to get something close to it. Because the demand had already been organically validated at pilot scale before the real product even launched, M-Pesa could scale into proven behavior rather than having to create demand from zero the way a from-scratch banking product would — which is a large part of why adoption reached 17 million accounts within five years.

the payoff

By 2012, roughly 17 million Kenyans held M-Pesa accounts, and by November 2014 M-Pesa transaction volume represented almost half the value of Kenya's entire GDP; by 2016 the platform was moving over KSh 15 billion (roughly $150 million) a day.

where it breaks

The signal only holds if the repurposing is genuinely widespread rather than a handful of atypical users — a pilot too small or too short to distinguish a real emergent pattern from noise doesn't justify redesigning the product around it. It also depends on the organization being willing to abandon the mandate it was actually funded or built for; a team too anchored to its original grant metric may suppress or simply not notice valuable off-label use because it doesn't serve the number they're measured on. And formalizing an improvised workaround into a real product still has to solve problems the informal version never faced — regulatory compliance, agent cash liquidity, fraud at scale — so a literal port of 'what people were already doing' can still fail even when the underlying demand signal was completely real.

what came after

M-Pesa became the reference case for mobile-money adoption worldwide, studied in a World Bank report and multiple Harvard Business School case submissions, and its model — build the tool around what pilot users actually do rather than what it was designed for — was replicated by mobile-money services launched across dozens of developing markets in Africa, Asia and Latin America in the years that followed.

references

1. [1]M-PesaWikipedia, 2024en.wikipedia.org
2. [2]The Making of M-PESASafaricom Newsroom, 2021newsroom.safaricom.co.ke
3. [3]Risk Sharing and Transactions Costs: Evidence from Kenya's Mobile Money RevolutionAmerican Economic Review, 2014aeaweb.org

keep it

same kind of clever

• 2000 · AIGA Design for DemocracyAfter Florida's 'butterfly ballot' cost thousands of voters their intended vote, the people who fixed it weren't election lawyers — they were typographers.
• 2001 · Richard Ankrom / CaltransAn artist tired of missing his exit built a Caltrans-perfect sign and installed it himself
• 2017 · Behavioural Economics Team of the Australian Government (BETA) / Australian Taxation OfficeAustralia's tax office cut inflated deduction claims by writing to the accountant, not the taxpayer