This note summarizes a research pilot prepared around Nepal's Hello Sarkar grievance system. The pilot uses 279 publicly released complaint records to demonstrate how a larger, privacy-protected dataset could support evidence-based public service improvement.
What the pilot is about
Hello Sarkar is a national grievance channel through which citizens submit complaints, suggestions, and inquiries about public bodies. The public dashboard shows more than 113,000 registered complaints, but most publicly visible information is aggregate. The pilot argues that record-level access, handled with strong privacy safeguards, could reveal far more useful policy evidence.
The released sample is too small for final national conclusions, but it is enough to show the value of the data structure. The records include complaint text, source channel, assigned office, complaint category, status, timestamps, action histories, public responses, and attachment signals.
Why it matters
Citizen grievances are not only isolated complaints. At scale, they can become a practical signal about where public services are failing, which issues repeat, how complaints move between institutions, and whether administrative closure reflects meaningful action.
- Citizen priorities: recurring concerns can show what people most often expect government to fix.
- Service bottlenecks: action histories can help identify delays, repeated forwarding, or weak follow-up.
- Early warning: sudden increases in complaints about roads, prices, water, health, transport, or documents can indicate emerging problems.
- Better routing: explainable AI can suggest responsible offices while showing similar past cases and human-readable reasons.
- Public trust: aggregate, privacy-preserving dashboards can show what government hears, forwards, and resolves.
What the 279-record sample suggests
The pilot sample points toward everyday service-delivery concerns: roads and transport, municipal services, payments and prices, health, water and sanitation, employment, education, and document-related services. It also shows that citizens use multiple channels, including website, Nagarik App, social media, national call, email, and physical or application-based submissions.
Because the sample is small, it should not be used to rank offices or draw strong national conclusions. Its value is methodological: it shows that the full dataset could support fair, normalized analysis by service domain, office, channel, geography, severity, status, and time period.
Proposed full-dataset study
The proposed study would request privacy-protected access to the full Hello Sarkar complaint dataset. Personally identifiable information could be removed or hashed before analysis. The work would focus on aggregate patterns, anonymized examples, and policy-relevant indicators.
- Complaint-domain taxonomy: build an auditable taxonomy for Nepali, English, and romanized Nepali complaint text.
- Policy dashboard prototype: show trends by concern domain, office, channel, time, and response stage.
- Explainable routing support: develop AI-assisted classification and office-routing suggestions with transparent reasoning.
- Response-quality analysis: distinguish forwarding, acknowledgement, follow-up, inspection, repair, payment, document issuance, and other substantive actions.
- Policy brief and research article: convert aggregate evidence into public-facing recommendations and peer-reviewed research.
Core idea
The full Hello Sarkar dataset could help Nepal move from complaint collection to public-administration intelligence. With proper privacy safeguards, grievance records can support faster routing, better planning, fairer monitoring, early warning, and more transparent evidence-based policy making.