Building Credit For Customers The System Has Not Yet Scored
FatakPay is a consumer fintech serving thin-file and low-CIBIL customers — borrowers with limited credit history who are structurally underserved by mainstream lending.
Fatak Udaan is its credit-building program, bundling a ₹500 Credit Builder Loan, an FD-backed lifetime-free SBM Novio credit card and score building through timely repayments, for a ₹299 + GST program fee. The product is affordable to start, but requires the customer to understand it before they will pay for it.
Interest Was Never The Problem. Understanding Was.
Users entered the Udaan journey and then abandoned it before completing the ₹299 + GST program fee payment. For thin-file customers, the decision was not price-sensitive — it was clarity-sensitive.
These customers needed to know whether the program would actually improve their loan approval chances, how the ₹500 Credit Builder Loan worked, what the fee covered and whether the transaction was secure. Without those answers, the journey stalled at the payment step.
| Friction Point | Observed Problem |
|---|---|
| Pre-Payment Drop-Off | 16,801 unique leads abandoned the Udaan journey before completing the program fee payment |
| Ambiguity | 3,189 users (30% of connected) never resolved into a clear yes or no |
| Deferral | 1,624 users (15% of connected) reached partial completion and asked to be called later |
| No Closure Channel | Intent captured on the call had no immediate payment mechanism to act on |
| Manual Data Loop | Daily Excel exchange of leads and outcomes limited same-day attribution and real-time retargeting |
Signal OS™ Separated Real Intent From Ambiguity.
Rather than treating every abandoned lead as an identical retry, SignalOS classified connected users by what they actually expressed on the call — separating genuine forward intent from deferral, disinterest and unresolved ambiguity, so that follow-up effort could be concentrated where it would convert.
| Signal | What It Revealed |
|---|---|
| Drop-Off Status | Where the user left the Udaan journey |
| Engagement State | Whether the user responded at all, or disconnected immediately |
| Payment Intent | Whether the customer expressed willingness to pay the program fee |
| Deferral Signal | Whether 'call later' or 'busy' masked genuine interest |
| Conversation Depth | AHT as a conversion proxy — 219s among paid users versus 90s overall |
| Payment-Stage Movement | Whether the nudge translated into observable journey progression |
| Architecture Layer | Deployment Capability |
|---|---|
| Customer Signals | Drop-off status • Engagement state • Stated payment intent • Deferral signal |
| Decision Intelligence | Intent classification • Forward-intent prioritisation • Rechurn eligibility |
| Journey Intelligence | TOFU drop-off recovery • Program education • Payment-stage progression |
| Execution Layer | AI Voice Agent (WhatsApp payment-link layer designed for expansion) |
| Human Intelligence | No live escalation layer in the July construct — the ambiguous pool was identified as the candidate for an assisted pass |
One Journey, Measured At Every Progression State.
The deployment ran a single core journey — same-day recovery of drop-off leads — but instrumented every state within it, so each stage of progression became measurable and separately addressable.
| Journey Stage | Trigger | Intervention | Outcome |
|---|---|---|---|
| Lead Dropped Off | Daily drop-off file shared till 11 am | Same-day AI outbound call | 10,672 connected (64%) |
| Connected & Engaged | User responds on call | Loan status informed; program explained; objections handled | 7,848 engaged (74% of connected) |
| Forward Intent | Positive or partial intent expressed | Intent tagged and shared back for follow-up | 3,676 forward-intent leads (34%) |
| Payment Stage | User proceeds post-nudge | Stage movement tracked via share-back | 1,131 moved to payment stage (10.6%) |
| Payment Completed | User pays ₹299 + GST | Success visibility fed back to Fundamento | 111 payments completed |
| Funnel Stage | Volume | % Of Unique | % Of Connected | AHT |
|---|---|---|---|---|
| Unique leads targeted | 16,801 | — | — | — |
| Connected | 10,672 | 64% | — | 90s |
| Engaged (user responded) | 7,848 | 47% | 74% | 122s |
| Forward intent | 3,676 | 22% | 34% | 168s |
| — Positive payment intent | 1,802 | 11% | 17% | 150s |
| — Partial completion + call later | 1,624 | 10% | 15% | 185s |
| — Partial completion + busy | 250 | 1% | 2% | 147s |
| Not interested | 983 | 6% | 9% | 160s |
| Ambiguous responses | 3,189 | 19% | 30% | — |
| Not engaged | 2,824 | 17% | 26% | 7s |
| Moved to payment stage | 1,131 | 6.73% | 10.60% | 136s |
| Paid | 111 | 0.66% | 1.04% | 219s |
One Voice Agent Carrying The Entire Education Load.
For a thin-file customer, the call is not a reminder — it is the explanation that makes the product make sense. The Voice Agent had to inform, educate, handle objections and capture intent inside a single conversation, without a human layer behind it.
| Agent / Layer | Responsibility |
|---|---|
| Voice Agent | Re-engages drop-off leads; informs loan and program status; explains the Credit Builder Loan, program fee and card benefits; handles objections; captures and tags payment intent |
| Workflow Layer | Orchestrates the daily data loop — lead intake, outcome tagging and share-back (Excel-based in the July construct) |
| WhatsApp Agent | Designed for expansion — payment-link nudge to forward-intent leads at the moment of captured intent |
| Routing Layer | Not in scope for the July construct |
Capabilities Are Not Projected. They Are Measured.
64% (10,672 of 16,801 unique leads)
Customer Connectivity
74% (7,848 users)
Engagement Among Connected
34% of connected (3,676 leads)
Forward Intent Captured
10.60% of connected (1,131 leads)
Payment-Stage Progression
111 payments (1.04% of connected)
Program Fee Payments Completed
219s AHT versus 90s overall
Conversation Depth Among Converters
From Drop-Off Calling To Closed-Loop Payment Recovery.
The July campaign established the funnel definition, the intent taxonomy and the conversion pool. The expansion plan builds directly on that intelligence — retargeting the forward-intent cohort with personalised scripts, and closing the gap between captured intent and completed payment.
- Forward-intent rechurn cohort — 3,676 connected users, split into positive intent, call later and busy
- Payment-stage movement as a leading indicator of conversion
- Intent-classification taxonomy enabling month-on-month funnel comparison
- Udaan VOC intent library — eligibility, Credit Builder Loan mechanics, fee and security, card and FD benefits
- Conversation-depth learning — scripts should invite questions rather than compress the pitch
| Started With | Expanded Into |
|---|---|
| Single-pass drop-off calling | Rechurn of the qualified forward-intent cohort |
| Generic re-engagement script | VOC-informed personalised scripts |
| Voice-only engagement | Voice plus WhatsApp payment-link closure |
| Manual Excel data loop | Real-time stage-movement visibility |
| In-month attribution | One-week conversion window so delayed payments are credited |
Four Questions Stood Between Interest And Payment.
| Customer Question | Business Impact |
|---|---|
| Will this actually improve my loan approval chances, and when can I reapply? | The core purchase driver for thin-file users — scripts must set honest eligibility expectations to convert without over-promising |
| How does the ₹500 Credit Builder Loan work — EMI, tenure, and how does it build my score? | An education gap on the product's core mechanism; a clearer explanation shortens the path to payment intent |
| What does the ₹299 + GST fee cover, and is the transaction secure? | A direct payment blocker — trust reassurance plus a payment link at the moment of intent addresses it |
| What do I get from the SBM Novio card and the FD — limit, rewards, requirements? | The strongest positive-framing lever, and the benefit-led hook for the personalised rechurn script |
