Product & engineering · v1 scope
The strategy, extracted for development
Six layers built around one advantage a wallet-only competitor cannot copy: the operator can lend — so good behaviour buys cheaper, faster credit, not just bigger discounts.
6
layers, ordered by trigger frequency
2 h
two separate holds shape the reward UX
25k
Rs./day Level 0 cap — confirmed
18 mo
four phases, Phase 1 in the first quarter
The one guardrail — the mercenary test
Extrinsic rewards spike in month one and fade by month three. So watch one pair of series, same cohort: Daily Habits and Moments engagement against Trust Score movement. Engagement up, Trust Score flat is reward-chasing, not habit formation — an instrumentation requirement from day one, not a launch-day report.
| 1 Credit is the currency, not just coupons — progression feeds a visible, weekly Trust Score. | 5 Give it a purpose bigger than discounts — literacy, small business, community. |
| 2 Reward the moment, not the month — instant and visible, never a multi-week draw. | 6 Design for redemption, not breakage — profit on unredeemed points is profit on disengagement. |
| 3 Keep the base layer boringly simple — real cashback, credited automatically, no ledger. | 7 Reward the KYC upgrade itself — front-load a bonus for clearing biometrics and Level 0. |
| 4 Fund it with partners wherever possible — brands already spending on acquisition. | 8 Make daily engagement free — zero-cost actions only, never a transaction to keep a streak alive. |
| Layer | What it is | Who funds it |
|---|---|---|
| Daily Habits | Streaks, fraud and literacy quizzes, light puzzles — all free actions. | Cheapest layer: small non-cash points redeemed into the partner marketplace. The fraud quiz is partly self-funding via avoided disputes. |
| Moments | Instant transaction-triggered mini-reward, max once a day, credited instantly. | Fixed marketing budget, capped like a CAC allocation — not an open-ended prize pool. |
| Foundation | Automatic real-time cashback, no points ledger. The default experience. | The bank's own margin, deliberately modest — its job is trust, not maximum value. |
| Trust Score | Weekly score from volume, bill punctuality and savings. Unlocks approval speed, limits, pricing. | Self-funding: lower cost-of-risk on lower-risk customers offsets the pricing concession. Needs credit-risk model integration. |
| Marketplace & Purpose | Voucher marketplace plus the purpose track — literacy, small business, community impact. | Merchant and brand partners paying for visibility and elevated earn rates. |
| Onramp | One-time bonus for biometric verification and clearing Level 0. | Fixed one-time cost per upgrade — budget it as customer acquisition. |
The deposit angle
ADB gating on the Root and Harvest tiers is a funding source, not a cost — every rupee pulled into savings and current accounts is low-cost retail funding, measured in basis points off blended cost of funds. Track it with Treasury, not Product alone.
| Tier | Qualifying behaviour | Foundation benefit | Trust Score unlock |
|---|---|---|---|
| Seed | Account opened, first three transactions | Welcome cashback credit | Baseline score established |
| Sprout | Regular monthly transacting | Elevated cashback on two to three types | Score visible; first pre-qualification signal |
| Root | Sustained 3 months+, on-time bills | Partner marketplace unlocked | Faster approval track; modest limit increase |
| Harvest | High consistent activity + savings held | Priority support; higher Moments rate | Preferential pricing on qualifying loans |
| Legacy | Top-decile 12 months+, or purpose milestones | Full marketplace and purpose benefits | Best available pricing and limits |
Names and thresholds are workshop material — the layering is what to protect.
Build requirement
One free streak freeze per calendar month. A single missed day must not reset progress.
Build requirement
No notification shaming a broken streak. One neutral reminder is fine; guilt-based copy is not.
Two to three questions a session, rotating, one to two points per correct answer, explanation shown after every answer. Starter bank — expand it, weighted to branchless-banking fraud:
| A caller from “your bank” asks you to read out the OTP you just received. | No — never over a call |
| A prize message asks for a small fee via QR code to claim it. | No — prizes never charge |
| An agent asks for your MPIN to complete a transaction for you. | No — agents never need it |
| Check the displayed name matches the merchant before scanning their QR. | Yes — QR swaps are real |
Track: quiz completion against downstream fraud-dispute rates by cohort. If it holds, report this layer to Risk and Finance as partly self-funding, not just as engagement. Games: Phase 4 only, tied to the purpose narrative — after the habit loop is proven.
| Constraint | What it means for the build |
|---|---|
| Cooldown A — funds received 2 hours, SBP |
Cash-outs, online purchases and top-ups against newly received wallet funds are held. A Moments reward is funds received: credit it visibly and instantly, but say it becomes spendable after the standard window. Never imply instant usability. |
| Cooldown B — security action 2 hours, all debits |
Triggered by registration, new-device registration, email or MPIN change — confirm the target MFB's equivalent. If the Onramp/BVS flow registers a device, expect a blanket debit freeze straight after: celebrate the bonus, but sequence the messaging. |
| Level 0 ceiling Rs. 25,000 / day |
Confirmed against the source terms. This is why the Onramp bonus exists: a Level 0 customer is structurally capped out of the rest of the program until they clear it. |
| Points liability IFRS 15 |
Model Daily Habits separately from Moments — many small pooling points versus occasional larger ones; one blended breakage rate will misstate it. Finance and Audit sign-off before launch. |
| Shariah framing if an Islamic window ships |
Frame rewards as hiba — discretionary generosity tied to real transactions — never a promised or benchmarked return. Chance mechanics work where the entry is a transaction happening anyway. Get named board sign-off on the mechanic, not the copy. |
Liability = Outstanding Points × (1 − Breakage Rate) × Cost Per Point
| Group | What to watch |
|---|---|
| Engagement headline |
DAU/MAU stickiness, weekly — the north star, and a churn signal two to three months ahead of revenue. Plus streak completion at day 7/14/30, quiz completion and correct-answer rate, and session frequency by cohort (one ratio can hide a power-user core). |
| Durability mercenary or real |
The mercenary test — engagement versus Trust Score movement, same cohort. Trust Score to repayment and default correlation against a control. Retention delta, participants versus non-participants, split by Daily-Habits-led or Trust-Score-led engagement. |
| Economics | Redemption rate sustained and rising — never minimised as a saving. Partner-funded share of reward value, rising. Points liability as a share of program revenue, modelled per layer, quarterly with Finance. |
| Risk | Mitigation |
|---|---|
| Becomes a discount-only clone, losing the credit link during build | Trust Score protected as a funded, tracked workstream from day one — not a Phase 2 item that slips |
| Regulatory or Shariah non-compliance in reward structuring | Compliance and the Shariah board in mechanic design, not just marketing review |
| Points liability understated — audit finding | Joint Finance and Audit liability model reviewed before launch |
| Loan margin cannibalised by over-generous Trust Score pricing | Pilot pricing and limits with a controlled cohort; risk-model validation as a launch gate |
| Execution quality undermines a sound design | Redemption reliability and support responsiveness as core KPIs from day one |
| One big-bang release delays time-to-market | Phase 1 ships on existing data, independent of the Trust Score work |
Out of scope for v1 — named so scope-creep has something to point back to
Games and puzzle mechanics (Phase 4, after the loop is proven) · the full Shariah variant (Phase 4, board sign-off before it starts) · a quantified financial model — exact NFI, exact bps on cost of funds — which needs the MFB's real cost-of-funds and adoption inputs, never invented placeholders · exact tier thresholds above Level 0, to be pulled from the live Schedule of Charges.
Onramp, foundation cashback, credit-linked trust score, instant moments, free daily habits, partner-funded purpose. The names are workshop material; the layering is the part worth protecting.