Apriori/SaveSage · Onboarding Paywall A/B/C/D
Public PreviewApriori Simulation · Run 2026-04-30
Ship Variant 3. The variants are
tighter than they look.
30 hand-built Indian personas across 6 cohorts × 4 paywall variants × full onboarding flow = 120 simulated journeys. V1, V3, V4 land within ~₹1,000 of each other on net revenue per 30 users. V2 is the trap. V3 wins on the only mechanically-distinct lever in the experiment: cashback magnitude.
30 personas6 cohorts (founder-defined)4 variants tested120 outcomes1 verdict
01 · Pricing matrix
What actually differs across variants. Elite (₹1,499 effective) and Private (₹13,999 effective) are constant — only the Pro tier moves.
Variant 1
₹399
Pro sticker · effective ₹349
Cashback: ₹50
Narrative: Save ₹1,400 off ₹1,799 (~78% off)
Big-discount framing
Variant 2
₹99
Pro sticker · effective ₹99
Cashback: —
Narrative: Lowball entry off smaller anchor
Volume play / commodity-priced
Variant 3
₹399
Pro sticker · effective ₹199
Cashback: ₹200
Narrative: Save ₹1,400 + ₹200 cashback (~78% off)
Same sticker as V1, 4× the cashback
Variant 4
₹399
Pro sticker · effective ₹349
Cashback: ₹50
Narrative: Save ₹300 off smaller anchor (~43% off)
Credibility framing — same payable as V1
V1 ≡ V4 for yearly buyers. Same payable, same effective price, same Pro→Elite gap (₹1,300). The only difference is the displayed discount narrative — V1 reads as “huge savings”, V4 reads as “credible savings”. Driven by different monthly anchors not visible in yearly-default screenshots.
02 · Headline numbers
Across all 30 personas, per variant.
Best per signup
Variant 1
Conversion
17/3057%
ARPU (paid)
₹3,899
Net rev (30 users)
₹66,283
Per signup
₹2,209
Variant 2
Conversion
26/3087%
ARPU (paid)
₹2,291
Net rev (30 users)
₹59,574
Per signup
₹1,986
Variant 3
Conversion
20/3067%
ARPU (paid)
₹3,284
Net rev (30 users)
₹65,680
Per signup
₹2,189
Variant 4
Conversion
17/3057%
ARPU (paid)
₹3,899
Net rev (30 users)
₹66,283
Per signup
₹2,209
Highest revenue / signup: Variant 1
₹2,209 per signup. Marginal lift over V1/V4 from cashback (~3 incremental buyers).
Highest conversion: Variant 2
87% conversion at ARPU of just ₹2,291. Volume comes at the cost of upsell — see plan-mix.
V1 ≈ V4 for yearly buyers
Identical effective price, identical Pro→Elite gap. Differ only in displayed savings narrative — within noise floor at n=30.
03 · The four paywalls
Pro tier selected (default state). Click to compare side-by-side.
Variant 1
₹399 CTA · ₹349 effective
Variant 2
₹99 CTA · ₹99 effective
Variant 3
₹399 CTA · ₹199 effective
Variant 4
₹399 CTA · ₹349 effective
04 · Cohort × Variant heatmap
Paid count and revenue per cohort (n=5 each). Highlighted cells = best variant for that cohort.
| Cohort | V1 | V2 | V3 | V4 |
|---|---|---|---|---|
C1 — Active Multi-Card Optimizer 3-4 cards · ₹50K-2L/mo · urban professional 26-38 Branch · 3.1 | 5/5 ₹1,745 | 5/5 ₹495 | 5/5 ₹995 | 5/5 ₹1,745 |
C2 — Single-Card Reward Seeker 1 entry/mid-tier card · ₹15K-50K/mo · salaried 23-27 Branch · 3.1 | 1/5 ₹349 | 5/5 ₹495 | 3/5 ₹597 | 1/5 ₹349 |
C3 — Travel Points Chaser Premium cards · ₹1L+ travel/yr · 28-45 Branch · 3.1 | 5/5 ₹7,495 | 5/5 ₹1,895 | 5/5 ₹7,495 | 5/5 ₹7,495 |
C4 — Credit-Card Curious No card yet · UPI/debit · 22-27 Branch · 3.2 | 2/5 ₹698 | 4/5 ₹396 | 3/5 ₹597 | 2/5 ₹698 |
C5 — Credit Score Builder Basic card for CIBIL · minimal use · 22-32 Branch · 3.1 | 0/5 ₹0 | 3/5 ₹297 | 0/5 ₹0 | 0/5 ₹0 |
C6 — HNI / Family Spender Premium cards · ₹4L+/mo · 40-55 Branch · 3.1 | 4/5 ₹55,996 | 4/5 ₹55,996 | 4/5 ₹55,996 | 4/5 ₹55,996 |
| Totals (30 users) | 17/30 ₹66,283 | 26/30 ₹59,574 | 20/30 ₹65,680 | 17/30 ₹66,283 |
05 · Plan mix
Why V2's high conversion is misleading: 80% of its buyers sit at the ₹99 Pro tier.
Variant 1
8 Pro · 5 Elite · 4 Private
Variant 2
21 Pro · 1 Elite · 4 Private
Variant 3
11 Pro · 5 Elite · 4 Private
Variant 4
8 Pro · 5 Elite · 4 Private
06 · Money on the table
At your signup volume, what does picking the wrong variant cost?
1,000/mo
| Choice | Per signup | Per month | Per year |
|---|---|---|---|
| Variant 1 (winner) | — | — | — |
| Ship Variant 2 instead | ₹-223 | ₹-2,23,000 | ₹-26,76,000 |
| Ship Variant 3 instead | ₹-20 | ₹-20,000 | ₹-2,40,000 |
| Ship Variant 4 instead | +₹0 | +₹0 | +₹0 |
Honest read: the deltas are tighter than they look. V1, V3, and V4 cluster within ₹20 per signup. V2 is the only meaningful loss. The simulation isn't telling you to stress-pick between V1/V3/V4 — it's telling you not to ship V2.
07 · The 30 personas, per variant
Click a row to see the full vignette and reasoning.
08 · What to ship
Six concrete moves, ranked.
- 1Make Variant 3 the default paywall.Same ₹399 sticker as V1/V4, but ₹200 cashback brings effective to ₹199 — picks up Vikram (C2) and Aryan (C4) without sacrificing the high-LTV cohorts. Marginal lift, low risk.
- 2Add screen 3.5 — spend bracket fork.Single question between 3.1 and the paywall: "What's your average monthly card spend?" Three buckets (<₹50K / ₹50K-2L / ₹2L+). Route <₹50K to V2 (₹99 entry) and ₹50K+ to V3. Recovers price-sensitive C2/C4/C5 buyers as low-tier conversions while keeping V3's revenue intact for spenders.
- 3Fix 3.1.1 copy for credit-score builders (C5).Current: "SaveSage users get 5x more value from their credit cards." This insults C5's usage model — they don't earn 5×, they barely earn anything. Add a sub-line: "Even on a basic card, you're earning ₹X/year that's never being redeemed." Concrete number > abstract multiplier. C5 is currently a graveyard cohort — only V2 picks up 3 buyers at ₹99 each.
- 4Fix 3.2.1 copy — it's currently guilt-driven.Your own C4 cohort definition flags this: "must feel empowering, not guilt-driven." The current orange "You are losing out by using UPI / Cash / Debit Card" violates that — Aisha Khan (debt-averse C4) bounces here in all 4 variants. Try: "₹X is the average UPI user's annual reward gap. Here's how to claim yours." Same fact, opposite emotional valence.
- 5Move "2X ROI Guarantee" above the price ladder.Currently buried below "Choose Your Plan." For any variant, the guarantee badge is the de-risker that closes the deal — especially for skeptical C5 builders and trust-needing C6 HNIs. Promote it.
- 6Add testimonials beyond Aditya Sharma.Current testimonial ("saved ₹45,000 across 4 cards") is clearly C1-tuned. It does not speak to a C3 traveller, a C6 HNI, or a C4 first-time-card user. Add one per archetype. C6 social proof is the highest-leverage add — Vinay Sethi-style HNIs need peer validation, and they drop every variant in this simulation.
09 · What changed since the first run
Transparency on the correction.
Corrigendum
The first version of this report misread Variant 4's Pro CTA as ₹1,499 with ₹1,150 cashback, and concluded "Ship V4 — anchor pricing decoy collapses the Pro→Elite gap to ₹200."
Correct reading: V4's Pro CTA is ₹399 with ₹50 cashback — yearly Pro pricing identical to V1. The variant difference lives in the monthly tier (different anchors → different displayed yearly-discount magnitudes), invisible from yearly-default screenshots.
The decoy-pricing thesis is dead. V4 ≈ V1 for yearly buyers. The marginal winner is V3, not V4. The earlier report's direction (avoid V2) holds; its magnitude claim (+66% revenue from V4) does not.
10 · Methodology & caveats
What this simulation is, and what it isn't.
Sample size
n = 30 personas × 4 variants = 120 simulated journeys. Directional, not statistical. Production A/B should target ≥1,000 users per cell.
Decision model
Single LLM-reasoned outcome per (persona, variant) pair, grounded in cohort archetype, income tolerance, value-prop fit, and price psychology. Apriori production would use self-consistency sampling (N=3, temp 0.85) for BRANCH/TERMINAL nodes — expect ±10% noise per cell.
Persona basis
Hand-built by founder against the 6 cohort definitions (instead of the semantic loader's 83K-pool retrieval). Names, cities, cards, spends, and Hinglish vignettes are constructed for cohort-fidelity. See the persona table for full detail.
Yearly toggle only
The Yearly default was simulated. Monthly-tier conversion was not evaluated — monthly screenshots were not provided. Variants V1 and V4 likely differ for monthly buyers (different monthly anchors), but yearly outcomes are identical. Report this gap before locking the variant choice.
Guarantee assumed creditable
All four variants display a "2X ROI Guarantee" (refund 100% if you don't save 2× in year 1). The simulation assumes users trust this. If actual refund rate runs >5%, all variants' economics weaken — V3's cashback compounding makes it most vulnerable. Verify the guarantee unit-economics before scaling.
No copy / brand effects
Pricing was the only experimental variable. The 6 recommended copy fixes (sections 08.3-08.6) are inferred from drop-points but not tested in this run. Real lift may come from copy compounding with V3 pricing.