Reconciling mobile-money payments with meter readings
In a rural water service, the money and the water are almost never described by the same file. Payments arrive over mobile money, timestamped to the second. Meter readings are noted on paper, then re-keyed. Subscribers live in a spreadsheet, water points in a GIS file. Each is internally consistent; together, they don’t talk to each other.
Reconciliation answers one simple question for every payment: which subscriber, which reading and which site does it belong to? Done by hand in Excel, it’s slow — and it lets through exactly what you’re looking for: the payments you can’t tie to anything.
What automatic reconciliation reveals
By linking a single month of files, every transaction gets classified:
- Reconciled — payment tied to a consistent subscriber, reading and site.
- Unmatched / orphan — no match found: potential revenue leakage.
- Duplicate — the same transaction counted twice.
- Wrong-site — payment recorded at a water point that isn’t the subscriber’s.
- Missing reading — a subscriber pays, but no reading was captured for the period.
Each category is quantified in € (or FCFA), by site and by agent. You no longer guess that “money is missing somewhere”: you show where, and how much.
Without transferring your data
Reconciliation doesn’t require exposing your subscribers. Names and numbers can be hashed before any processing, and the analysis can run entirely on your machine — one month of files is enough. You keep your data; you get the read-out.
Methodological honesty: reconciliation is a diagnostic on the files provided, not an audited financial statement. An “unmatched” amount can reflect a missing file as much as a true leak — it’s a starting point to verify, not a verdict.