evidoria

← Back to browse

Good practice Imported

WASAC's Smart-Metering and Machine-Learning Pilot Cuts Non-Revenue Water by 23% in Kigali, Rwanda

Rwanda · Kigali · See the Rwanda profile · See the Kigali profile

Evidence: Observational / pre–post Top 62% 53/100 · Ask Evidence Copilot about this practice

Rwanda's state water utility WASAC piloted smart meters in Kigali and worked with researchers applying machine-learning predictive modelling to nearly a decade of network data, cutting non-revenue water by 23% within six months against a historical rate of 35–45%.

23%
Non-revenue water reduction (6-month pilot)
38%
Customers reducing water use
41%
Drop in billing complaints
60+
WASAC staff trained
50,000
Additional digital water meters (Itron)
110,000 units
Prior digital meter rollout

Details

Maturity
Pilot
Promoter
WASAC Ltd (Water and Sanitation Corporation, Rwanda)
Period
2014–2024
Keywords
water utilities, public infrastructure, environment

Context

Non-revenue water has historically run at 35-45% for WASAC, Rwanda's state water utility, well above the government's 25% target.

Activities

A smart-metering pilot in Kigali trained over 60 WASAC technicians, billing agents and customer-support staff; a companion study applied panel-data analysis and machine-learning predictive models to WASAC's network data spanning July 2014 to June 2023.

Results

NRW fell by 23% within six months of the pilot, 38% of participating customers reduced their water use, and billing complaints dropped 41%.

Conclusions

The pilot and modelling work are complementary rather than a single integrated AI product; the 23% result is an early, localised outcome rather than a nationwide, sustained figure.

Implementation
Indicative cost
Medium (€50k–€500k) — Combines a Kigali smart-metering pilot with an academic ML modelling study; hardware costs include a 50,000-unit digital meter contract with Itron on top of an earlier 110,000-unit rollout.
Time to results
Medium (1–3 years) — Pilot NRW result achieved within 6 months; ML study covers July 2014-June 2023 data, published June 2024.
Staffing & skills
60+ WASAC technicians, billing agents and customer-support staff

Conditions for success

  • nearly a decade of historical network data (2014-2023) available for ML modelling
  • smart-meter hardware rollout (Itron) alongside the pilot

Common failure modes

  • result is an early, localised 6-month outcome, not a nationwide sustained figure
  • historical NRW baseline of 35-45% still well above the 25% government target

Commonly funded by

National / regional programmes

Indicative funding routes for practices of this type — always check each programme's current calls and eligibility rules.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful