Customers are told we can install, but technicians arrive to find locations unserviceable — wasting capacity and breaking trust.
Data window: Feb 25 – Mar 27, 2026 | Source: Snowflake prod_db | 2,766 paid bookings analyzed
Problems are baked into the sequence. Money is collected before address validation, and partners cannot verify coverage before assignment.
The current flow collects payment at step 4 but doesn't validate the actual address and partner coverage until step 10. By then, we've made a promise we may not be able to keep. The customer has paid, expects installation, and the partner discovers the location is outside their network.
Four root causes identified from data
The original problem: "technicians arrive to find locations unserviceable." Data shows that's only 15% of the story.
Partner reads address, rejects on paper before anyone dispatched. Shows up as "delay in installation" to customer.
Rohit dispatched then declined. Wasted tech visit. Of these, 57 (22%) later installed by another partner.
Key Finding
The matching logic itself needs overhaul, not just a pre-dispatch gate. 85% of failures are caught before dispatch — but only because partners reject bookings Genie shouldn't have sent them.
GPS disagreement predicts overall cancel rates (39% → 72%) but does NOT predict address-specific declines — those are flat at ~16% regardless.
| Source disagreement | Cancel Rate |
|---|---|
| Exact match | 2.8% cancel |
| <50m | 41.0% cancel |
| 50-200m | 44.7% cancel |
| 200m-3km | 55.0% cancel |
Worst case: GPS #1 shows Delhi, GPS #2 shows Bihar (864km), chat address says "Block D, New Ashok Nagar, Delhi" — three sources, three different stories, no reconciliation
Even with address and lat/long, partners rely on phone calls to confirm location. They haven't understood the effort Wiom takes to match — and have no stake in making it work.
"Sorry sir, us side hamari line nahi hai"
— Cable doesn't go there
"वहां तो भाई लग नहीं पाएगा हमारा"
— Understood address perfectly, infra gap
"society में permission नहीं होती है"
— Gated society restriction
Key Finding
Address quality is IDENTICAL for declined vs installed. Avg length 51 chars vs 50 chars. Partners use "can't understand address" as proxy for "my cable doesn't reach."
Genie finds the right AREA — but sends to the wrong PARTNER.
| Partner A declined → Partner B installed | Times in 30 days |
|---|---|
| Bharat Broadband → Bhatia Cable TV | 3 times |
| M R NETWORKS → Bottam Cable TV | 3 times |
| TECHFIBER → Umang Cable TV | 3 times |
22 partner-cell combos had the same partner decline 2+ bookings in the same ~100m area. 136 total reroutes where A declined and B installed — none fed back into routing.
84% of bookings are within 25m of a known base point — Genie finds the right AREA. But 8.8% still get partner decline because it sends to the wrong PARTNER. Dense multi-partner areas have the highest decline rate (9.3%).
Once we accept a booking, we MUST install. No false promises — payment only after partner confirms serviceability.
Partner sees their own customers on a map around the booking. Data they can't ignore — not just a text address.
Every install expands green zones. Every decline sharpens boundaries. Every reroute fixes partner routing. The system gets smarter with each booking.
Partners are stakeholders in zone accuracy — bigger green zone = more guaranteed leads. Decline in committed zone = financial recovery. Good behavior rewarded, bad behavior self-corrects.
Don't take money until serviceability is confirmed by the partner. Three zones, built from data, confirmed by partners, honest to customers.
70% of bookings
5+ customers in ~100m cell from ANY partner combination. Area is proven serviceable. Best-performing partner gets priority with fair rotation.
Customer pays immediately — area is proven
Routed to best scorer, rotation for fairness
No monopoly — partners compete on performance
21% of bookings
1-4 customers in cell — some evidence but not proven. Broadcast to partners. Partner confirms → then payment taken.
Payment ONLY after partner confirms
2-hour SLA for partner response
No confirm = "Not available yet"
10% of bookings
No customer data in area. Customer told honestly. No false promise, no payment.
No payment taken, no false promise
Customer notified when area expands
Never routed until partner expands
The zone system naturally rewards good partners and sidelines bad ones — no manual intervention needed.
Installs consistently → green zone grows → more auto-confirmed leads, guaranteed revenue
Frequent declines → green zone stays small → only amber broadcasts, must compete for each lead
Declines in committed zone → zone shrinks + financial recovery → fewer leads, trust score drops
We tested different green zone definitions against actual booking outcomes.
| Green Zone Definition | Coverage | Install Rate | Addr Decline |
|---|---|---|---|
| GREEN: 5+ customers (area-level) | 70% | 44.1% install | 22.2% decline |
| — Single-partner cells within green | 14% | 35.3% install | 7.4% decline |
| — Multi-partner cells within green | 55% | 46.3% install | 25.9% decline |
Key Finding
Area-level green (5+ customers) gives 70% coverage with 44.1% install rate. Within green, multi-partner cells install MORE (46.3%) but waste 3.5x the effort on address declines (25.9% vs 7.4%) — proving the problem is routing, not area coverage. Layer 2 (performance-based routing) eliminates this wasted effort without creating monopolies.
Customer gives location, we check with real partner data, and tell the truth. Payment only when a partner is committed.
Map View
Apna ghar ka address search karein
B-42, Sector 15, Noida, UP 201301
🏠 Sahi location dena zaroori hai — hum isi location pe partner check karenge
Partner: Sharma Cable Network
Install within 2 days
Payment sirf confirmed booking ke liye
Hum local partner se confirm kar rahe hain...
Aapko 2 ghante mein pata chalega
Koi payment abhi nahi lagega
Agar partner confirm kare → hum aapko notify karenge
Abhi aapke area mein available nahi hai
Jaise hi expand karein, aapko notify karenge
Koi payment nahi liya gaya
Aapko 2 ghante mein pata chalega
Partner ko check karne mein thoda waqt lagega
Koi payment abhi nahi lagega
Agar partner confirm kare → hum aapko notify karenge aur tab payment hoga
Agar koi partner confirm nahi karta → hum aapko bata denge "abhi available nahi hai"
Green zones auto-created from install history. Amber bookings broadcast for confirmation. No partner drawing polygons — just confirming what data already shows.
Wiom ne aapke install data se yeh area suggest kiya hai
Is zone mein aap decline nahi kar sakte. Booking aate hi install karna hoga.
47
installs
0
declines
Priya Singh
H-12, Block C, Mayur Vihar
Aapka 1 customer 200m mein hai
Kya reason hai?
Customer ko abhi payment nahi liya gaya hai. Aapke confirm karne ke baad hi booking hogi.
100m mein aapke 8 customer hain
Rajesh Kumar
3rd Floor, B-42, Sector 15, Noida
Yeh aapke core zone mein hai. Decline available nahi hai.
Green zones grow organically from partner behavior, not declarations. Every install and every decline reshapes the coverage map.
Every install adds a confirmed point. Amber cells near it graduate to green.
Decline + structured reason = that cell is marked for THIS partner. Other partners still get it.
Next time a booking hits this cell, Partner B gets priority. Partner A doesn't see it.
No partner can service here. Customer told "not available" upfront. No wasted time.
Conservative targets based on root cause analysis. These numbers compound as the self-learning model matures.
| Metric | Today | After |
|---|---|---|
| False promise (paid but can't install) | ~46% of bookings | 0% by design |
| Booking → Install rate | 37% | 70%+ (confirmed only) |
| Wasted tech visits (post-assign decline) | 89 in 30 days | < 10 |
| Booking cycling (pre-assign decline) | 524 in 30 days | Eliminated |
| Time to first install | Days of cycling | Green: instant. Amber: 2hr max. |
| Partner trust score accuracy | None | Auto-built from 262K installs + 136 reroute signals |
Pre-Dispatch Serviceability — Product Strategy Document
March 2026
What if we could eliminate partner matching entirely and turn every Wiom router into both a serviceability proof AND a zero-cost acquisition channel?
Today we ask: "Which partner should we send this booking to?" — and the partner says no 26% of the time.
The right question is: "How do we KNOW — with certainty — that Partner X's cable reaches Building Y, without asking Partner X?"
Every solution that still involves asking the partner is just a fancier way of asking. What if you never need to ask?
262K routers are already proving serviceability 24/7
Every Wiom netbox broadcasts a "Wiom Net" SSID. That's 262K routers constantly proving where cable infrastructure exists. Each router belongs to a specific partner. If a new customer's phone detects the signal — the cable is physically there.
Customer opens app at home. App scans WiFi in background.
Detects "Wiom_Net_XXXX" nearby. That router = Partner A, installed 30m away.
Cable physically exists. Auto-assign Partner A. No GPS, no address, no partner call needed.
Why it kills every RCA
RCA1: No decline possible — cable proven. RCA2: GPS/address irrelevant. RCA3: Partner doesn't need to confirm. RCA4: No Genie routing — physics routes.
Self-learning built in
Every new install = new broadcasting router = bigger signal footprint. Coverage grows automatically. Partner's own hardware defines their territory.
Turn every router into a zero-cost billboard that only targets serviceable addresses
What if the WiFi signal doesn't just verify serviceability — it generates demand? Rename the public SSID to "Ghar ka Internet". Every non-Wiom person in range sees it on their phone. They tap to connect. Captive portal opens.
Person sees WiFi
"Ghar ka Internet" appears on their phone at home
Taps to connect
Captive portal: "Unlimited home WiFi from Rs500/month"
Drops their info
Name, phone, flat number. Gets 15 min free WiFi as reward.
Pre-qualified lead
Within signal range = 100% serviceable. Partner already known.
Zero CAC
No ads, no pamphlets, no door-knocking. 262K routers already broadcasting 24/7. Every router is a billboard that costs nothing extra.
100% Serviceable Leads
If they can see the signal, cable reaches them. No serviceability check. No partner confirmation. No decline. Book and install.
Self-Reinforcing Loop
Every new customer adds a new "Ghar ka Internet" signal. More installs = more signals = more leads = more installs. Compounding growth.
The math: 262K routers x ~20 neighbors in signal range = 5M+ impressions/day at Rs0 CAC
Even a 0.1% conversion = 5,000 pre-qualified leads/day. Each one guaranteed serviceable, partner pre-matched, zero matching friction.
Zone Framework (Primary Solution)
Classifies areas, gates payment, routes to best partner. Still relies on historical data + partner confirmation for amber zones. Incremental improvement on existing flow.
"Ghar ka Internet" (Alternate)
Eliminates matching entirely. Physics proves serviceability. Same signal generates demand. No GPS, no address, no Genie, no partner call. The infrastructure IS the product AND the marketing.
Additional approaches that use community, economics, and visual proof to solve matching from different angles.
Ask the nearest existing Wiom customer: "Aapke padosi ko Wiom chahiye. Kya aapke partner ki cable unke building tak pahunchti hai?" Neighbor gets Rs50 credit on install.
Broadcast booking to nearby partners: "Deposit Rs200 to claim. Install in 48h or lose it." Only confident partners bid. Market reveals truth. Repeat decliners self-eliminate.
Customer sends 30-sec video of their building/gali. Partner watches, recognizes the area, confirms or declines — BEFORE payment, BEFORE dispatch. Kills 776 "can't understand address" events.
Rohit walks his cable route with GPS on. 2 hours per partner, one-time. Creates true cable-path graph. Match by "can cable reach this building?" not "is this building CLOSE to a point?"
Partner publishes capacity: "I can add 5 customers on Gali X this week." Customer sees "instant install available." Like Swiggy Instamart — supply before demand. Zero declines.
Pay nearby customer Rs100 to walk to new customer's building and check cable visibility. Costs Rs8,900 vs Rs26,700 in wasted Rohit visits. 262K scouts on every gali.