What the average M-Pesa agent earns per month
This is not an estimate. Safaricom publishes both halves of the sum in its own full-year results, so the average can be calculated rather than guessed.
| Year to 31 March | FY2026 | FY2025 | Change |
|---|---|---|---|
| M-Pesa commissions paid (KSh m) | 37,375.4 | 37,223.1 | +0.4% |
| Number of M-Pesa agents | 333,011 | 298,890 | +11.4% |
| Average per agent, per year | KSh 112,235 | KSh 124,538 | -9.9% |
| Average per agent, per month | KSh 9,353 | KSh 10,378 | -9.9% |
Both input figures come from Safaricom PLC's FY26 Results Booklet published on 7 May 2026: the Safaricom Kenya direct costs table on page 24, and the Safaricom Kenya operating KPIs on page 12. The per-agent rows are ours, and they are simply the first line divided by the second. Checked against the booklet on 29 September 2026.
Two caveats worth stating before you use that number
- It is a ceiling, not a wage. "M-Pesa commissions" is Safaricom's total commission cost for the channel. Divided by agents it gives the average commission attributable to an outlet before rent, staff, tax and any split with an aggregator. Nobody takes KSh 9,353 home.
- The average hides a very wide spread. A stand outside a matatu stage and a stand in a quiet estate both count as one agent. Reported figures of KSh 20,000 to KSh 100,000 a month are not necessarily wrong, they are the top of a distribution whose middle is this low.
You may see KSh 112,244 a year quoted elsewhere for the same period, against our KSh 112,235. That version divides the Safaricom Group commission line by the Kenya agent count. The gap is KSh 9 a year, because Ethiopia's M-Pesa commission was KSh 3.2 million against Kenya's KSh 37.4 billion. Either figure is fine. We use Kenya on both sides of the sum because it is the cleaner comparison, not because the other one is wrong.
Why the average is falling
The commission pool did not shrink. It grew, by 0.4%. What changed is the number of people dividing it: 34,121 more agents joined the network in twelve months, an 11.4% increase. More shops splitting the same money is the whole story, and it is why the average fell 9.9% in a year when Safaricom actually paid out more. Any plan to open an outlet should assume the denominator keeps growing.
How agent commission actually works
Safaricom pays on a banded schedule, by transaction value, on both the deposit and the withdrawal leg. Because it is banded and not a percentage, there is no single rate: a shop doing many small deposits and a shop doing a few large withdrawals earn very different blended rates on the same volume.
That is why the calculator asks for your blended commission percentage rather than assuming one. Read it off your own statement: take last month's total commission, divide by last month's total value processed, and enter that. If you are modelling a shop you do not run yet, ask an agent in a similar location for theirs, and sanity check the result against the KSh 9,353 average above.
What comes off before you see it
- Withholding tax. Agency commission is withheld at source. KRA publishes the resident rate for consultancy and agency, in force since 1 July 2003, as 5%. It is a credit against your final tax, not an extra tax, but it is cash you do not hold in the meantime.
- The aggregator split. If you operate under a principal agent rather than directly, your contract decides how the gross commission is divided. Splits of 70/30 and 80/20 in the outlet's favour are commonly described, but Safaricom does not publish a standard and we have not verified one. Read your own contract. This single term can move your take-home more than anything else on this page.
- Turnover tax. At 1.5% of monthly turnover for businesses with annual turnover between KSh 1 million and KSh 25 million, under section 12C of the Income Tax Act as amended by the Finance Act 2023.
- Rent, staff, power and security. The calculator has fields for these because they are usually the difference between a shop that clears and a shop that does not.
The float economics
Float is working capital, and the right question is not "how much does the shop make" but "does this beat leaving the same money somewhere else". The calculator returns an annualised return on float for exactly that comparison. Put your SACCO's expected dividend or a money market fund's current yield next to it before you commit capital.
Where it goes wrong
- Float runs dry. A customer wants a 50,000 withdrawal, you do not have it, they walk to the next shop and often do not come back. Keep float at roughly three times your largest typical single transaction.
- Theft and fraud. USSD scams and fake reversal calls target agents specifically. Verify every reversal directly with the customer, never off the back of a call.
- Location decline. A stand losing footfall, through a mall closure or road works, cuts revenue immediately while rent stays exactly where it was.
- More agents nearby. The 11.4% growth above is not evenly spread. In a dense trading centre it lands as two new shops on your street.