A cleanliness reporting and accountability system for Faridabad — designed around a city that generates 900 tonnes of waste a day and processes less than a third of it.
Faridabad is the largest city in Haryana by population and one of the ten lowest-scoring cities in the 2025 Swachh Survekshan. Its municipal corporation has spent roughly a hundred crore rupees a year on sanitation since 2020 and the streets have not changed much. That gap between spend and outcome is what I wanted to design against.
I picked the city deliberately. I live thirty kilometres from it, the data is public and unflattering, and the failure is one of coordination and visibility rather than of funding, which is the kind of problem an interface can actually move.
What follows is the full arc: the secondary research, the service map of what happens today when a resident reports a garbage heap, the three user groups the system has to serve at once, and the flows and screens I designed for them.
Every figure in this case study is from a public source and cited at the end. The personas are composites built from secondary research, and the usability protocol is a study I designed but have not yet run. I have kept the two clearly separated rather than dressing up the second as findings.
A city of 24 lakh people, 46 wards and one landfill that stopped taking its rubbish in September 2025.
Bandhwari, commissioned in 2008 for 1,200 tonnes a day and now 19 metres high, stopped accepting Faridabad's waste after 30 September 2025. A waste-to-coal unit at Mothuka, 20 km out, is still only a proposal.
About 2,900 sanitation workers are deployed against the 7,000 routine operations need. Sanctioned strength (3,600) was last updated over two decades ago when the city was half this size.
Two segregation plants at Mujheri and Pratapgarh cost ₹2+ crore. Three other sites were abandoned after protests. More than 60% of the city's daily waste still leaves untreated.
Comparison of daily waste generation, installed capacity, actual processing, and untreated residue.
Trommel machines separate what arrives; they cannot separate what was never sorted. Mixed waste entering a segregation plant comes out as low-value reject, which is why capacity sits idle. Household segregation is the input the ₹2 crore plant needs to run at all. The product must earn its keep at the doorstep hours before the truck arrives — making the household, not the ticket queue, the real design surface.
Workers deployed vs corporation required strength across civic zones.
A product that adds a form to a safai karamchari's morning will be abandoned by week two. The worker-facing side has to take steps away. It must replace the paper muster roll and the supervisor's morning phone round, not stack new friction on top of a 400-house beat that is already impossible to finish.
Faridabad has spent over a hundred crore rupees on sanitation in four years, has two processing plants and a national ranking that keeps falling. Something between the budget and the street is disconnected.
Attendance is on paper. Route completion is a supervisor's phone call. Nobody can answer "which of the 46 wards was actually swept this morning" without ringing three people.
There is an MCF-311 app and a toll-free line. A complaint goes in and nothing comes back — no worker name, no photograph, no closure the resident can dispute. People stop reporting.
Households are told to sort into green and blue. Nobody tells them what happened to yesterday's sorted bag. Compliance collapses the first time someone watches a truck mix them back.
How might we make the waste system visible to the three groups who move it — the household that sorts it, the worker who lifts it, the inspector who is answerable for it — so that each one can see the result of what the other two did?
Eight steps, three handoffs, and no point at which the resident and worker ever see each other.
Mapping Sunita's citizen reporting emotional curve and Ramkishan's shift reality across critical touchpoints.
| Stage | 1. Discover & Trigger | 2. Channel & Report | 3. In-Progress & Work | 4. Verification / Closure | 5. Long-Term Retention |
|---|---|---|---|---|---|
|
Citizen (Sunita) Reporting Flow |
Action: Steps out at 7:10 AM, spots overflowing heap near D-Block park. 😠 Frustrated "This heap has been sitting here since Monday." |
Action: Opens Safai Setu, taps giant camera, auto-locks GPS in 15s. 😐 Cautious "Didn't need to type an essay or make an account." |
Action: Receives live ticket ID (FBD-14-2291) with 11:00 AM SLA countdown. 🙂 Reassured "Assigned to Ramkishan on Beat 14-B. There's a timer." |
Action: Notification at 10:24 AM with side-by-side after photo. Taps 'हाँ, साफ़ है'. 😃 Delighted "I saw the actual photo proof. It's really gone." |
Action: Checks 30-day streak & Ward 14 weekly rank. Reports again next time. 🌟 Empowered "My reports actually change my neighbourhood." |
|
Worker (Ramkishan) Field Shift Flow |
Action: 6:00 AM beat start. Opens phone; offline queue active. 😐 Focused "Beat 14-B: 412 houses on my round today." |
Action: Ticks lanes of 12 houses with 1 tap. Logs locked gate in 56dp grid. 🙂 Relieved "No typing with dirty gloves. Just 4 big icon taps." |
Action: Sees assigned heap on D-Block park with resident's before photo. 😐 Clear Goal "Knows exact GPS spot without supervisor calling." |
Action: Clears heap, snaps 1 after-photo. Ticket sent for resident review. 😃 Dignified "Auto-stamped proof of my hard work on record." |
Action: Finishes shift at 9:48 AM; attendance automatically marked on muster roll. 🌟 Valued "Saved a 2 km walk to the municipal store to sign paper." |
|
System Impact Design Opportunity |
Opportunity: Remove login barrier; infer ward & beat from geo-coordinates. | Opportunity: 10-second voice note fallback for non-Hindi typing residents. | Opportunity: Public SLA clock visible to both resident & inspector. | Opportunity: Binary resident verdict + 6h auto-closure window. | Opportunity: Ward-level civic pride replaces abstract city-wide shaming. |
Separating what secondary research settled from what requires primary fieldwork.
MCF filings, tender notices, NGT orders on Bandhwari, Swachh Survekshan scorecards, and 4 years of Haryana reporting. Where sources disagreed, I kept the range rather than the flattering number.
The version a solo designer can actually finish without a corporate research budget:
Composites built from secondary research — labelled that way deliberately as the first thing to update after fieldwork.
Her Day: Out by 7:10 AM. Bin cycle arrives at 8 AM. She leaves the bag with the security guard and never sees collection happen.
Her Phone: 3-year-old Redmi, 4G that drops in stairwell, WhatsApp & PhonePe power user, Hindi keyboard.
Tried Before: Toll-free number twice, RWA WhatsApp group 4 times, tweeted photo once.
His Day: 6 AM start. A beat of ~400 houses sized for two people and worked by one. Detours to store to sign paper muster roll.
His Phone: Entry-level Android, 16 GB with storage permanently full. Reads icons/numbers fast, long Hindi text slowly.
Frictions: Locked houses, overflowing bins at 6 AM, supervisor calling repeatedly to ask where he is.
Her Day: 14 wards, 3 WhatsApp groups, and an evening paper register filled at 6 PM from memory and phone calls.
Her KPI: Measured purely on complaints marked closed, causing closure without resolution.
Blind Spot: Cannot distinguish an understaffed ward from a poorly run one, so cannot advocate for resources.
The heap — or the bin, or the missed house — is the record. Each surface writes a different part of it, and every part is visible to the other two.
If reporting takes longer than a WhatsApp photo, nobody uses it. The camera is the front door.
A named person with a photo builds human accountability that a status badge never will.
Designed for workers who navigate numbers and visual iconography faster than long sentences.
Offline queue under 12MB. Weekly ward competition beats abstract national city-shaming.
Hindi by default · 1-tap reporting
Complete navigational taxonomy and user flow hierarchy across all three operational surfaces.
Before committing to high-fidelity visual UI, core interaction hypotheses were stress-tested using low-fidelity grayscale wireframes focused on cognitive friction, touch ergonomics, and zero-typing workflows.
Rapid 10-minute pencil sketches testing gesture economy, no-keyboard data logging, and bin visual metaphors before drafting digital wireframes.
Analyzing one-handed ergonomic reachability for on-the-go reporting alongside real-time civic operational density across Faridabad's 46 wards.
Single-handed natural thumb interaction on 5.5"–6.5" Android screens while walking or holding brooms.
Supervisor real-time telemetry: Live heap resolution time vs. workforce deficit by ward.
High-fidelity interactive mobile UI specifications designed for real-world municipal conditions in Haryana.
Three screens, one of them optional. Location and category are inferred and shown as editable chips; no account required before first report.
Duplicates within 40m merge into "6 people reported this" to add weight rather than rejecting.
A ticket cannot reach "closed" without an after-photo, and the resident gets the last word. Rejection escalates to the Sanitary Inspector.
Reopen rate replaces closure rate. Rejection never penalizes the worker directly, avoiding gamed numbers.
Drag the central slider handle to compare the Before vs After photo evidence submitted by field worker Ramkishan.
The worker app has no keyboard, no login screen after first setup, and no category picker. It exists to replace the paper muster roll and the supervisor's phone call.
Finishing the round marks him present, which is the one piece of paperwork that currently costs him a detour.
This is the flow that feeds the ₹2 crore of idle plant capacity. A household that never learns what happened to its sorted bag has no reason to sort the next one.
"9 of 14 houses on your lane are on a streak" is social pressure at a scale where people know each other.
Eleven components, two type families, and a high-contrast palette that survives direct Haryana sunlight and a scratched screen. Nothing here is decorative.
Green carries the citizen surface; orange carries the worker surface. A glance across a lane identifies the user role. Red is strictly reserved for reopened tickets.
Devanagari runs one optical step larger than Latin to maintain equal x-height legibility in direct sunlight. Never below 13px.
48dp standard touch targets; 56dp on worker exception grid for single-handed use with rubber gloves. Two transitions only (<180ms).
The After-Photo. Even though it is the most expensive element in bandwidth and worker field time, without it this is just the app the city already has. It is the single structural change that alters the civic accountability service rather than just decorating an interface.
Written as a test protocol rather than as fake findings. The three assumptions below are the ones that would sink the product if wrong.
If the answer is no, round data never arrives and the dashboard is blank. Test: Shadow 6 shifts with a clickable prototype, measure abandonments, ask at the end of the round.
The loop-closing argument rests on this. Test: 2 cohorts of 30 households over 6 weeks (one receiving after-photos, one receiving status changes). Measure second reports.
The only reason plants would ever run above 50% capacity. Test: 1 lane of 14 houses graded daily for 8 weeks against an un-graded control lane.
4 residents across zones, 3 safai karamcharis, 1 sanitary inspector. Tested on their own phones to test device performance reality.
A pilot in one ward for one quarter. Four numbers, none of them vanity engagement metrics.
| Measure | Today | After One Quarter | Why This Metric |
|---|---|---|---|
| Segregated Share of Collected Waste | Baseline (~30%) | +15 pts | The plants are the constraint. This is the only number that turns idle capacity into processed tonnage at the vehicle. |
| Median Time from Report to Verified Clear | Unknown | < 6 hours | The city currently cannot answer this. Establishing a baseline is itself a civic transparency breakthrough. |
| Reopen Rate on Closed Tickets | Not tracked | < 10% | Near-100% closure with unmeasured reopen rates proves a system is being gamed. This is the honest accountability metric. |
| Households Reporting a Second Time | Baseline | > 40% | Repeat reporting is the ultimate proxy for citizen trust. The fastest early signal whether the loop is alive or dead. |
Honest reflections on where the process fell short and what holds up under scrutiny.
Six weeks of secondary reading gave me a defensible picture of the waste infrastructure, but told me little about human nuances. The personas are composites. If I ran this again, I would spend week one walking a beat with Ramkishan before opening Figma.
I drew resident screens first because that is what portfolio audiences expect. But the citizen app is worthless without ground data, and round data depends entirely on a worker finding the tool genuinely helpful.
Streaks and leaderboards are easy to draw and easy to pitch. Whether a family in Sector 21C cares that Lane 3 is beating Lane 4 is unknown, and the tax rebate requires a municipal resolution nobody has proposed.
Faridabad's waste problem will be solved by procurement, land acquisition, and hiring, not by an app. What an app can do is make the failure visible to the people who can act on it, which is a smaller claim and a more honest one.
All figures retrieved September 2026 from published municipal and press records.