Prepaid water only works when tokens sold and water dispensed agree. We watch the gap.

Kenya pioneered the water ATM — smart-card kiosks dispensing affordable water in Nairobi's informal settlements since 2015, topped up over M-Pesa. The model works; the operational question is trust: is every dispensing point up, is dispensed volume matching token sales, and is anyone bypassing the meter? addanode is not a prepaid meter vendor — we're the independent telemetry layer that verifies the fleet: uptime, flow-vs-token reconciliation, tamper alerts and one dashboard across every kiosk and prepaid zone. (Planning or costing a vending operation instead? Start with our water ATM business guide and 2026 machine price guide.)

What this solves in Kenya

The three gaps in every prepaid water scheme.

A dead kiosk is invisible until the queue forms

A water ATM that's jammed, empty or powerless earns nothing and burns community trust — and with kiosks scattered across estates and settlements, the operator usually hears about it from complaints. Uptime telemetry (power, valve, tank level, dispensing activity) turns a dead point into a same-morning repair ticket instead of a lost week of sales.

Token sales ≠ delivered water

The payment platform knows what was sold; only an independent flow measurement knows what physically left the pipe. A gap between the two is theft, a failing meter, a leaking valve or a software fault — all expensive, all invisible if the only record is the vending system grading its own homework.

Bypasses and tampering erode the model

Prepaid schemes concentrate their losses at the meter: magnetic interference, plumbing around the meter, valve forcing. Independent flow sensing upstream of the dispensing point — compared continuously against vended volumes — makes a bypass show up as data, not as a year-end revenue mystery.

Local operating constraints

Fitted to how prepaid water actually runs in Kenya.

  • The ecosystem is mature — we join it, we don't replace it. Kenya's water ATM model (utility partnerships with kiosk hardware providers, NFC prepaid cards, M-Pesa top-ups) has run for a decade. Our sensors and gateways sit alongside whatever vending hardware you operate, reading flow, pressure, power and tank level independently of the vendor's own reporting.
  • Kiosk sites are hard sites. Informal-settlement and market locations mean vandal exposure, unreliable power and no fixed internet. Hardware is enclosed and tamper-alarmed, runs on solar/battery, and reports over 4G or LoRaWAN — with local buffering through outages.
  • Reconciliation needs your vending data. Flow-vs-token comparison works best when vended volumes flow into the dashboard too — via CSV export or API from your vending platform. Where that's not possible yet, independent flow trending alone still exposes dead points, night leaks and bypass patterns.
  • County and donor reporting is part of the job. Many Kenyan kiosk fleets answer to a county WSP, an NGO funder or both. Delivered-litres records per point, exportable and timestamped, are the difference between an anecdote and a verified impact report.

Public references here — the Nairobi water ATM programmes and their NFC/M-Pesa model — are drawn from published reporting. Our verification pattern comes from metering and telemetry work across Southern Africa, adapted to kiosk fleets.

Typical deployment

What a verification layer looks like on a kiosk fleet.

1 · Instrument the points

Independent flow sensing on each kiosk's supply, tank level where storage exists, power status, enclosure tamper switch. Ten minutes of plumbing per point, no change to the vending hardware.

2 · Connect the fleet

Each point reports on its own link (4G/LoRaWAN); the fleet appears as one map on addaNet with per-point status, litres dispensed and last-seen time.

3 · Reconcile & alert

Vended-vs-measured comparison per point per day; alerts on dead points, abnormal night flow, tamper events and points selling but not dispensing (or dispensing but not selling).

4 · Report

Monthly delivered-litres and uptime per point — the record your county partner, funder or board actually wants to see, exported instead of compiled.

A pilot on 5–10 dispensing points is typically live in 2–4 weeks, hardware pre-configured in Johannesburg, installed by your technicians under remote guidance — same platform as our NRW monitoring and water telemetry in Kenya.

The three gaps, and the instrument that closes each

Gap in the schemeWhy it opens in KenyaInstrument that closes it
Vend records vs water deliveredM-Pesa tokens exist; delivered-volume records rarely doContinuous delivered volume per kiosk or zone
Bypass and tamperingPrepaid-electricity tampering skills already existBulk-vs-sum-of-units per zone
Dry tank behind a vending meterToken systems keep selling when the tank is emptyTank level and pump state on the vend platform
Who this serves

Prepaid operators we build for in Kenya.

Water ATM operators and their utility partners — kiosk fleets in Nairobi and secondary towns where uptime and reconciliation decide both revenue and reputation · County WSPs running prepaid zones — where prepaid was deployed to fix collections, and the remaining question is whether the meters themselves are honest · NGO and development-funded schemes — community water points whose funders want delivered-litres evidence, not estimates · Estate and private suppliers — bulk-buying operators reselling through prepaid points, where the buy-vs-sell balance is the whole margin.

FAQ

Prepaid water verification in Kenya — common questions

Do you sell prepaid water meters or water ATMs?

No — deliberately. Kenya has established kiosk and prepaid-meter suppliers; what the market lacks is an independent layer that verifies the fleet is up, honest and reconciled. Vendor-neutrality is the point: our numbers are worth something precisely because we don't sell the meters being measured.

How does flow-vs-token reconciliation work?

An independent flow sensor measures what physically passes to the dispensing point. Your vending platform reports what was sold. The dashboard compares the two per point per day: a persistent gap flags theft, meter drift, leakage or a software fault — each with a different signature. Without the independent measurement, the vending system is auditing itself.

Can you monitor kiosks without mains power or internet?

Yes — that describes most kiosk sites. Sensors run on battery or small solar; connectivity is 4G or private LoRaWAN depending on the site; and every device buffers locally so a power cut or network outage never leaves a hole in the dispensing record.

We run a prepaid zone, not kiosks. Does this apply?

Yes. For prepaid household zones, the same principle applies one level up: an independent zone bulk meter reconciles total prepaid vends against actual zone consumption, exposing bypasses and failing meters in aggregate — and telling you which streets deserve a physical inspection. It pairs naturally with DMA-based NRW monitoring.

What does it cost per dispensing point?

Instrumentation per point is modest — a flow sensor, small controller and connectivity — and it's quoted as a written band after a free remote review of your fleet list and vending platform. A single recovered bypass or a week of avoided kiosk downtime typically covers a point's monitoring for a long time.

Does it integrate with M-Pesa top-up records?

We don't touch the payment rails themselves — reconciliation uses vended-volume records from your vending platform (export or API), which is where M-Pesa top-ups already land as tokens. That keeps the verification layer independent of payment processing while still closing the loop from shillings to litres.

Know every kiosk is up — and every litre is paid for.

Send us your fleet list and vending platform. We'll propose a pilot on your busiest points, with uptime, reconciliation and tamper alerts on one dashboard.