M-PESA: Choose the Next Use #
A payment can succeed on a phone and still fail the person who needs the money. This exercise uses Behavior Market Fit to ask which complete action is worth supporting, for whom, and under what conditions.
Your task: recommend the next bounded investment after a mobile-money pilot. Compare a loan-repayment service, a person-to-person transfer service, and an extension that resolves a specific uncertainty. Submit a decision and a test that could overturn it.
Status: public-source teaching reconstruction, version 1.0. The decision role, option menu, prompts and proposed tests are original instructional material. The historical participants are not known to have used DRIVE, BMF or BFA. This exercise does not establish that those methods improve decisions.
The decision setting #
Imagine a product team reviewing the M-PESA pilot in Kenya during 2006. Its task is to select a first commercially supportable service. The team can improve the current workflow, test a different use, or defer expansion while resolving a named uncertainty. It cannot assume unlimited agent capacity, free distribution or ongoing project-team rescue.
This is an invented decision meeting, not a recovered board paper. The factual packet draws on accounts published around or after the pilot. It limits the exercise to pilot-stage observations, but it cannot reproduce everything a real decision maker knew at one date. The actual next steps and later outcomes are on the separate worked page.
Choose a direction before opening that page. If you already know M-PESA’s history, write down what you remember and set it aside. Assess the options using the packet’s evidence and its gaps.
The public evidence packet #
A. The service and its early limits #
A WRI chapter in Vodafone’s Economic empowerment through mobile, reviewing the 2006 setting, describes a loan-repayment pilot with Faulu, Safaricom and Commercial Bank of Africa. Customers could also transfer money, buy airtime and exchange cash with agents. The report describes reduced repayment travel, but also limited agent access and withdrawal constraints. It says use beyond repayment was slow and commercial acceptance remained unresolved because phones and service were free. Its relevant notes cite an April 2006 pilot review. BS-0079
B. What the participant account adds #
Hughes and Lonie’s 2007 account dates the pilot start to 11 October 2005 and its official end to 1 May 2006. Nearly 500 clients and eight agent stores began the pilot. Selection favored phone familiarity; clients received phones and account credit. Training covered menus, PINs and transactions. Agents needed reassurance and cash floats. Reported uses included business payments, remote family support and repayment for other people. Faulu faced reconciliation difficulties. Agent electronic balances could also run out. These are participant-reported observations, without a complete opportunity-to-completion dataset. BS-0079
C. What you cannot conclude yet #
The accounts offer different emphases. Examples of an alternative use can coexist with slow adoption of that use across a pilot. Neither fact supplies a frequency distribution. We cannot calculate how many eligible people completed a remittance, how many returned at their next need, or what each successful delivery cost under ordinary operating support.
Treat the observed alternative uses as candidates to investigate. Treat the original repayment use as a candidate too. The packet does not justify a numeric BFA score, a paid-demand estimate, or a claim that all Kenyans face the same conditions.
Your choice #
The following alternatives are teaching constructions. You may amend them, but keep the same business objective and explain what your amendment changes.
| Option | Candidate action and population | Work the service would have to support | Most important evidence request |
|---|---|---|---|
| Improve loan repayment | A borrower makes a due payment that reaches the correct loan record | Borrower access, collection, confirmation, group arrangements and institutional reconciliation | Are payments correctly credited with less total work for borrowers and the institution? |
| Test person-to-person transfer | A sender with a genuine transfer need sends money that the intended recipient can access | Onboarding, cash conversion, correct recipient details, transfer, cash availability and error recovery | Does the full transfer work repeatedly at acceptable total cost, including both ends? |
| Extend the pilot before choosing | Keep expansion bounded while observing a named weak link in an existing route | A defined observation protocol, an operating owner and a decision deadline | Which missing observation would distinguish the leading options? |
Include a less-user-work alternative in your comparison. For example, ask whether a service could remove a reconciliation task or a separate visit instead of asking every borrower to learn another payment sequence. This is a design hypothesis. Specify who would take over the work, with what authority and cost. Moving a burden to an agent or employee does not make it disappear.
Use workbook steps 1 and 2 to separate the desired outcome, the observable action and the proposed feature.
Follow the money all the way through #
Build an action chain for the option you favor. The table below is an instructor’s analysis checklist, not a claim that every listed failure occurred in this pilot.
| Actor | Action or responsibility to check | Evidence to request |
|---|---|---|
| Sender or borrower | Obtain usable access; understand the fee; enter the intended payment; protect credentials | Observation of the real task, including a declined attempt and any assistance |
| Cash-in agent | Accept and verify cash; provide the correct electronic value | Matched records; waiting time; available balance; staff time |
| Platform and customer support | Route the payment; confirm its status; resolve exceptions | Reconciled transaction records; delay and recovery records |
| Recipient and cash-out agent | Confirm the right person and amount; make funds usable when needed | Recipient confirmation; cash availability; travel, waiting and failed visits |
| Group treasurer or delegated payer, if used | Match each person’s payment to the group’s payment without losing the individual allocation | Member-level records matched to the group payment and final loan credit |
| Lending institution, if relevant | Credit the correct borrower; reconcile collection and settlement | Loan record matched to payment; exceptions; staff time and ownership |
| Bank and service finance team | Keep customer funds and system balances accounted for | Settlement and reconciliation evidence reviewed by the responsible operator |
| Agent manager and service owner | Train staff, maintain cash and electronic balances, fund support and own unresolved problems | A costed operating plan that ordinary staff can execute |
Where does your action start? What observable event finishes it? Mark every necessary handoff. For a transfer intended to supply cash, a sent message is an intermediate event. For repayment, the lender’s correct credit is part of completion.
Use workbook step 3. Label each row as a historical observation, your inference, or a proposal that still needs evidence.
Submit a decision that can be challenged #
Complete the decision workbook. Your answer must contain:
- Decision and population. Name the next investment, the people it serves and the complete action it should enable. State what remains outside the initial population.
- Evidence comparison. Give the strongest reason for your choice and the strongest reason against it. Explain how you interpret the two source accounts without averaging them into a success rate.
- Fit and support. Separate a durable reason to make the payment from the skills, access and support needed to execute it. Assign an owner and cost to each necessary support.
- A realistic test. Specify genuine opportunities, first completion, repeat completion, the observation window, fees and ordinary assistance. Include the recipient or institution that finishes the chain.
- Decision rules. Set proposed continue, revise, stop and insufficient-evidence conditions before seeing results. Explain why the thresholds would matter to the business and to users.
- Reversal condition. State which result would make you choose the strongest alternative. If your choice is to extend the pilot, name the evidence that ends the extension.
Use steps 4 through 8 to complete the test and next decision. Any numbers you choose are proposed planning rules. They are not historical M-PESA results or universal BMF thresholds.
Source chronology and reading boundary #
| Source | Events or observations covered | Publication and use in this exercise |
|---|---|---|
| WRI chapter in Economic empowerment through mobile, printed pp. 12-13, PDF pp. 14-15 | Pilot-stage evidence; relevant notes cite an April 2006 review | Reviews the 2006 setting; exact publication day is not established here. It is not an April decision document. |
| Hughes and Lonie, M-PESA: Mobile Money for the Unbanked, printed pp. 70, 72-78 | Pilot and subsequent development | Published in Innovations, Winter & Spring 2007. A participant-authored retrospective; the packet uses its pilot material. The full paper reveals later decisions. |
Sources checked 5 September 2026. External publications retain their own rights. The original teaching material is covered by the teaching-pack reuse terms.
After you record your choice: open M-PESA: Worked Decision to compare two reasoned responses, a proposed test and the subsequent history.