The verdict in three sentences
Collecting across multiple wallets multiplies statements, formats and errors: without tooling, the typical unexplained variance hits 1.5 to 3 % of volume. The 2026 fix is automatic reference-based reconciliation, matching each wallet collection to an order and a bank payout. Target outcome: under 0.5 % variance and reconciliation time cut tenfold.
Why multi-wallet breaks the books
In Kampala, a merchant typically accepts M-Pesa and Airtel Money. Each operator provides a CSV export with its own columns, its own identifiers and its own settlement timing. Manual reconciliation means aligning three sources: orders, wallet notifications and received transfers.
| Data source | Format | Key identifier | Availability delay |
|---|---|---|---|
| Store orders | Internal DB | order_ref | Real time |
| M-Pesa | CSV export | transaction_id | T+1 |
| Airtel Money | CSV export | payment_ref | T+1 |
| Bank statement | Statement / API | transfer label | T+1 to T+2 |
The trap: a customer pays UGX 90,000 on M-Pesa, the payout arrives net of fees (~1 %) at UGX 89,100, with a bank label that drops the order reference. Without automatic matching, the gap compounds.
The automatic reconciliation method
The principle: one unique reference key injected into every payment, then a three-pass match.
| Matching pass | Criterion | Target resolution rate |
|---|---|---|
| Pass 1 — exact reference | order_ref = payment_ref | 85 to 92 % |
| Pass 2 — amount + date | net amount ± fees, same day | +5 to 8 % |
| Pass 3 — manual review | remaining gaps flagged | < 3 % |
| Final target variance | unexplained | < 0.5 % |
With this approach, most items resolve automatically and only the residue needs a human eye. Operator fees are computed and provisioned line by line, so you never confuse "fees" with "cash shortfall".
Mini case study
David runs a hardware store in Kampala and collects UGX 30,000,000/month: 55 % M-Pesa, 45 % Airtel. Before tooling he spent 6 h/week (24 h/month) reconciling, with an average unexplained variance of 2 %, or UGX 600,000/month of unjustifiable shortfalls. After reference-based reconciliation: 30 min/week (2 h/month) and variance down to 0.4 %, or UGX 120,000. Gain: 22 h/month saved and UGX 480,000/month of gaps explained — over UGX 5.7 million/year.
Need a professional website?
Kolonell builds websites that attract clients, optimized for the Sénégalese market. Free quote in 2 minutes.
FAQ
Why do I have gaps when every payment went through?
Because payouts arrive net of operator fees and at T+1, with bank labels that drop your references. Without matching, these timing shifts look like cash shortfalls.
How long does manual multi-wallet reconciliation take?
Expect 4 to 6 h/week for a merchant active on two wallets, versus roughly 30 minutes with automated reference-based tooling.
What variance level is acceptable?
A healthy 2026 target is under 0.5 % of unexplained volume. Above 1.5 %, you need to structure reconciliation.
Do I need an API or is a CSV export enough?
Per-operator CSV export is enough to start. A bank or aggregator API automates imports and further shortens close time.
Does this work with more than two wallets?
Yes: the same reference key and matching passes apply to M-Pesa, Airtel, MTN or an aggregator without changing the logic.
Let's talk about your project. We automate your M-Pesa + Airtel multi-wallet reconciliation with a sub-0.5 % variance target. WhatsApp +221 77 596 93 33.
Mohamed Bah
Fondateur, Kolonell
Passionate about digital and entrepreneurship in Africa, Mohamed has been helping Sénégalese businesses with their digital transformation since 2020. Founder of Kolonell, he believes every SME deserves a professional and accessible online présence.
