Back to blog

Automation

M-Pesa Reconciliation: Automate Payment Matching and Reminders

How Kenyan businesses can stop ticking off M-Pesa messages by hand, know who has really paid, and follow up without the awkward phone calls.

Author
By OKWIYA Technologies
Published
Reading time
8 min read

A customer sends you a screenshot and says, "Nimelipa." You scroll through your M-Pesa messages looking for the amount. There are three payments of the same amount this morning. One has a name that does not match anyone you know. Your attendant thinks one of them was for yesterday's order.

Ten minutes later you are still not sure who has paid.

If this sounds familiar, the problem is not M-Pesa. M-Pesa did its job. The problem is that the record of the payment and the record of the sale live in two different places, and a person is the only link between them. That link is slow, it gets tired, and it goes home at 6 PM.

This guide explains why M-Pesa matching goes wrong, what you can fix this week for free, and what full automation looks like when you are ready for it.

What M-Pesa Reconciliation Really Means

Reconciliation is a big word for a simple question: does the money that came in match the sales you recorded, and do you know which payment belongs to which customer?

For a kiosk with a few sales a day, your memory might be enough. Once you have staff, credit customers, deposits, part payments or more than one till, memory stops working. You start finding out about mistakes at the end of the month, when it is too late to do anything about them.

Why Payments Stop Adding Up

Most mismatches come from a short list of causes:

  • Customer errors. Someone pays the wrong till, or types the wrong account number on a Paybill.
  • Duplicates and overpayments. A customer pays twice, or sends more than the bill, and nobody records the difference.
  • Reversals nobody wrote down. A staff member reverses a payment and the books never hear about it.
  • Timing. A payment made close to midnight can land on a different day's statement from the sale.
  • Wrong tagging at the till. A cash sale gets rung up as M-Pesa, or the other way round.
  • Fake confirmations. A forwarded or edited message is treated as proof of payment.

There is one more problem that is getting harder to work around. Customer names and phone numbers shown to merchants are increasingly masked, so matching on a name is no longer reliable. The transaction code, the amount and the time are what you should trust.

Start Here: Free Fixes for This Week

You do not need new software to get most of the way to clean records. Start with these habits:

  • Never release goods on a screenshot. Confirm the payment in your own M-Pesa messages or statement before anything leaves the shop.
  • Record the M-Pesa transaction code against every sale. That code is unique. A name is not, and an amount certainly is not.
  • Use account numbers if you have a Paybill. Ask customers to enter their invoice number, house number or phone number as the account. It gives every payment a label.
  • Do a daily close, at the same time each day. Compare the day's M-Pesa total with the day's recorded sales. A difference found today takes minutes to solve. A difference found in a month may never be solved.
  • Download your statement regularly and keep it in one shared Google Drive or cloud folder, so the records do not depend on one phone.

The daily close is the single most useful habit on this list. Even if you never automate anything, doing it will show you where money is leaking.

Paybill or Till: Why It Matters for Matching

The way customers pay you decides how easy matching can be.

A Paybill lets the customer enter an account number along with the payment. If that number is an invoice or a customer ID, a system can match the payment automatically and correctly most of the time.

A Till (Buy Goods) has no account number field. That makes it perfect for quick over-the-counter sales, but harder to match to a specific order after the fact.

If you take payments for orders, invoices, rent or school fees, a Paybill or an STK push (the prompt that appears on the customer's phone) usually makes life easier. If you sell over the counter, a Till is fine, as long as each sale is recorded when it happens.

What Automated M-Pesa Matching Looks Like

Safaricom's Daraja API lets your own system receive a notification each time a payment reaches your Paybill or Till. You register a web address with Safaricom, and details of each payment are sent there, including the transaction code, the amount, the time and the account reference the customer entered.

In plain terms, here is what changes:

  • A customer pays, and the payment appears in your system within moments, with nobody typing anything.
  • The system looks at the account reference or the STK push request and marks the right invoice or order as paid.
  • The customer gets an automatic confirmation.
  • Anything the system cannot match goes into a short "needs checking" list for a person to review.

That last point matters. Good automation does not pretend to be perfect. It handles the routine cases and puts the odd ones in front of you, so instead of checking every payment you check the few that need you.

STK push makes this even cleaner, because your system starts the payment itself and already knows which order it belongs to. The customer just enters their PIN.

Automatic Payment Reminders That Do Not Annoy People

Once the system knows who has paid, it also knows who has not. That is what makes reminders easy, and it is why matching should come first. Nothing damages a customer relationship faster than chasing someone who paid last week.

A simple reminder sequence looks like this:

  • A friendly note a few days before the due date
  • A reminder on the due date, with your Paybill details and their account number
  • A firmer but polite message if the payment is late
  • A flag to you, so a person makes the call on the difficult cases

You can send these by SMS or WhatsApp. If you use the official WhatsApp Business Platform, Meta now charges per template message, and a utility message, such as a payment update, is free when it is sent within an open 24-hour customer service window. Outside that window, it is charged. Plan your costs around that.

Only send reminders to people who are actually your customers and who expect to hear from you, and always make it easy for them to reach a real person.

Where AI Fits In

AI is useful here, but it is not the part that should decide whether money has arrived. The matching itself should be plain rules based on M-Pesa data. You want the same correct answer every time, not a clever guess.

AI earns its place around the edges:

  • Writing reminder messages in a tone that suits your customers, in English or Swahili
  • Answering "Have you received my payment?" on WhatsApp by checking your records
  • Suggesting a likely match for an unmatched payment, for a person to confirm
  • Summarising the week: who is overdue, how much, and for how long

Who Should Automate First

You do not need to automate because it is fashionable. It is worth doing when:

  • You or a staff member spend real time every day checking payments
  • You sell on credit, take deposits or collect monthly payments, such as rent, school fees or subscriptions
  • You have more than one branch, till or staff member receiving money
  • Customers have been asked to pay for something they already paid for

If you are selling a few items a day for cash, the daily close and a good notebook may be all you need for now. There is no shame in that.

How Okwiyatech Can Help

At OKWIYA Technologies, we build the link between your payments and your records so that you are not the link any more. Depending on your business, that could be:

  • A custom system with M-Pesa payments matched to orders or invoices and automatic reminders (okwiyatech.co.ke/solutions/custom-software)
  • WhatsApp setup for confirmations and payment updates (okwiyatech.co.ke/solutions/whatsapp-solutions)
  • An AI assistant that helps answer customers and handle routine admin (okwiyatech.co.ke/products/ai-secretary)
  • A ready-made product, such as our Property Management System for landlords and property managers (okwiyatech.co.ke/products)

The best next step is our free webinar, AI & Automation for Kenyan Businesses, on Saturday 21 November 2026, from 10:00 to 11:00 AM EAT on Google Meet. In one hour, we will show you how to use AI and WhatsApp to save time and how to stop chasing payments. No tech skills needed, and you can join from your phone.

Register free at okwiyatech.co.ke/webinar

Prefer to talk first? Chat with us on WhatsApp on +254 741 273 216, or book a free consultation at okwiyatech.co.ke/solutions/consultation.

Sources

  • Meta, Pricing on the WhatsApp Business Platform: https://developers.facebook.com/docs/whatsapp/pricing
  • M-Pesa Daraja C2B integration guide, DEV Community: https://dev.to/msnmongare/m-pesa-daraja-api-c2b-integration-guide-djn
  • Paybill KE, M-Pesa reconciliation mismatch guide: https://paybillke.com/guides/mpesa-reconciliation-mismatch-2026
  • Tally Solutions, Common payment reconciliation problems in Kenya: https://tallysolutions.com/ssa/business-guides/common-payment-reconciliation-problems-kenya/