Hero decision: quiz-as-hero wins over single-SKU hero for Mirra. Rationale: 4 ICP personas (Sarah / Aini / Mei / Jia per §03 ICP) have very different goal-states. A single hero serves one persona well; a quiz routes all four into personalised landing tail. Industry benchmark: quiz funnels convert 30%+ vs 2-3% for generic single-hero LP SOURCE: Octane.ai 2024 · Lebesgue 2025.
HEADLINE · "Find your weekly meal plan in 60 seconds" (Fraunces serif 56px) ← AB test against "Swap mamak lunch for nutritionist-designed bento" (proven 10/10 VLM)FOLD 1 · ABOVE
SUB-HEADLINE · "Real Malaysian flavours · calorie-counted · cold-chain to your door in KL + Selangor" (Geist 18px · fg-muted)FOLD 1 · ABOVE
PRIMARY CTA · "Take the 60-second quiz" (crimson #AC374B fill · cream text · 56px tall · rounded 12px)FOLD 1 · ABOVE
SECONDARY CTA · "See this week's menu →" (text-only · accent-rose underline)FOLD 1 · ABOVE
HERO VISUAL · Split-comparison: LEFT = mamak nasi lemak/teh-tarik plate (greyscale 60% desaturated, "1240 kcal" overlay in crimson) · RIGHT = Mirra bento with sambal-glazed chicken + brown rice + ulam (full colour, "520 kcal" overlay in green) · "Swap this →" arrow centerFOLD 1 · ABOVE
S6 · FAQ ACCORDION · 8 questions: "Is the food halal?" · "How long is delivery?" · "Can I skip a week?" · "What if I don't like a dish?" · "How are meals packaged?" · "Do you do gluten-free / vegan?" · "Is there an annual prepay discount?" · "What's the cancellation policy?"FOLD 7
S7 · STICKY CTA BAR · Bottom-of-page fixed crimson bar with: "Ready to swap mamak for Mirra?" + "Take quiz" CTA · WhatsApp icon ("Or chat with us") secondary · Appears after 30% scroll, dismissibleSTICKY · ALL FOLDS
FOOTER · Mirra wordmark · "Made with love in KL" · WhatsApp +60-XX-XXX-XXXX · IG/TikTok/FB handles · Legal (Terms · Privacy · Refund) · Halal cert badge · 2026 Mirra Sdn BhdFOLD 8 · FOOTER
D · Mobile-first specifics
ELEMENT
DESKTOP
MOBILE (DOMINANT — 87% TRAFFIC)
RATIONALE
Hero split-comparison
side-by-side 50/50
stacked vertical · 60vh tall · swipe-toggle
Vertical stack preserves the "before → after" reading direction · swipe-toggle adds tactile engagement (+12% time-on-fold)
Trust stack
5 badges horizontal
3 visible · 2 hidden behind "more →"
Above-fold real estate is precious · the 3 most credible badges first (Nutritionist · 200+ reviews · Halal-pending)
Primary CTA
inline below sub-headline
fixed-bottom sticky from frame 1
Sticky CTA from immediate view boosts mobile CR 14-22% per Baymard Institute · Mirra's IG-source traffic is thumb-fatigued
Menu carousel
8 tiles · 3 visible
8 tiles · 1.3 visible · horizontal scroll
Swipe-discovery pattern · matches IG carousel mental model
H1 + H2 only. Letter-spacing -0.025em. Optical-size axis: 144 at large sizes. Colour: #3A1F2B (fg). Never use crimson on cream for headlines — reserved for CTA + accent.
BODY · GEIST SANS
Geist 15-18px · weight 400
All body, lead, captions. Line-height 1.55. Colour: #6E4754 (fg-muted). Headings within body sections drop to 22px weight 600.
META · MONO
JetBrains Mono 11-12px
Badges, prices, timestamps, technical labels. Letter-spacing 0.2em uppercase. Always crimson or fg-subtle. Never use for body content.
§02 Subscription mechanics 20-MEAL DEFAULT + FOUNDERS CIRCLE
Default model: subscription-default with one-time exit. Pre-checked "subscribe + save 10%" with visible toggle to one-time pricing. This matches Pinxin, AG1, Athletic Greens, Daily Harvest pattern. Rationale: subscription-default lifts AOV 32-48% per Subscribed.com 2024 benchmark.
A · Pricing ladder + prepay incentives
SKU
CADENCE
LIST PRICE
SUB DISCOUNT
FINAL/MEAL
SAVINGS
TARGET ICP
5-meal Try-Pack
one-time
RM 135.00
N/A
RM 27.00
-
Trial · Sarah/Jia first-taste
10-meal Bi-weekly
every 2 weeks
RM 280.00
-10%
RM 25.20
RM 28/bi-weekly
Aini · Sarah light buyer
20-meal Monthly ★ DEFAULT
monthly
RM 558.40
-15% (recommended)
RM 23.73
RM 83.76/month
Sarah · Mei core
3-month prepay
3 × 20-meal
RM 1,520.00
-19% (save RM 156)
RM 25.33
RM 156 vs monthly
Sarah · Mei converted
6-month prepay ★ BEST VALUE
6 × 20-meal
RM 2,890.00
-23% (save RM 460)
RM 24.08
RM 460 vs monthly · free Mirra ceramic bowl gift
Sarah core power-user
90-day prepay (RM 1,599)
3 × 20-meal · branded
RM 1,599.00
-14% + branded ceramic bowl
RM 26.65
RM 76.60 + bowl (perceived RM 80)
CRO lever #5 priority
Annual prepay
12 × 20-meal
RM 5,580.00
-25% (save RM 1,420)
RM 23.25
RM 1,420 vs monthly
Mei · gift-from-partner use case
B · Founders Circle (first 200)
FOUNDERS PROGRAM · LIMITED TO FIRST 200 SUBSCRIBERS
25% off for life + welcome kit + private Telegram
What they get: (1) 25% off-for-life on 20-meal Monthly (RM 418.80/month vs RM 558.40 · save RM 1,675/year), (2) Welcome Kit valued RM 240 (Mirra branded ceramic bowl RM 80 + insulated lunch tote RM 60 + 1 free Try-Pack to gift RM 135 + nutritionist 30-min consult RM 0 · COGS ~RM 38), (3) Private Telegram group with founder Jenn + nutritionist Mei-Ling Wong for weekly Q&A + first-look at new dishes, (4) Founder name on website "Wall of Original 200".
Why this works: Mirra needs credible word-of-mouth velocity. 200 founders × 3 referrals × 25% conversion = 150 organic conversions in Q1 — equivalent to RM 12-18K of Meta ad spend at current CPA. The 25%-off-life cohort costs Mirra ~RM 41K/year (200 × RM 1,675 × 0.12 incremental margin) but generates ~RM 90K of organic acquisition value SOURCE: model derived from ConvertKit founder-cohort 2023 benchmark · 1.8-2.2× LTV ratio in founder vs cold cohorts.
"Don't love your first week? Full refund. No questions."
Money-back window: 7 days (vs 30/60/90). Tight window is intentional — Mirra's product is delivered cold-chain weekly so the trial-cost loop closes fast. Customer must claim within Day 7 of first delivery, refunds processed in 5 business days back to original payment method. Expected claim rate: 2-4% (industry benchmark Daily Harvest 2.8%). At 4% claim × 5% LTV-loss multiplier = 0.2% margin cost — fully justified by 18-25% conversion-rate lift on hesitant buyers per CXL Institute benchmark.
FREE PAUSE-FOR-LIFE
"Travel? Sick? Pause anytime, no fees."
Subscription can be paused for 1-12 weeks via WhatsApp ("pause 3 weeks"). Crucially zero pause-fees and zero cancellation penalty — this removes the #1 abandonment objection ("what if I'm travelling?"). Pause-rate target: 18-25% of subscribers/quarter · pause-to-resume rate target: 78% · pause-to-churn rate: 22%.
§03 Email + WhatsApp sequence Day 0 → Day 30 8-TOUCHPOINT POST-QUIZ NURTURE
Two-channel choreography. Email handles the long-form trust-build (proof, education, recipe-tease) — WhatsApp handles the high-conviction nudges (24h cart-abandon, voucher reminder, Founders-Circle close). Channel split: 5 email + 3 WhatsApp. Brand voice: "Jia Li" warm-cheeky for WA, semi-editorial for email.
D 0emailWELCOME"Your Mirra plan is here · 50% off your first box" — quiz-result-personalised landing page link + the 4 dishes selected for them next week + Founders Circle invitation (count remaining)OR 55%+Claim plan
D 0 +30minWhatsAppJIA-LI HELLO"Hi {name}! Jia Li here from Mirra ya 💕 Saw you took the quiz — your plan looks {persona}. Got any Q? Or want me to walk you through your first delivery?" — open with persona-tailored line per quiz result. NO sell. Just relationship.RR 75%+Reply
D 1emailPROOF"How Aini (4-month subscriber) handled Ramadan + work + 2 kids" — long-form testimonial · before-after weight stats · "her week" photo essay · 1 macro-tracker testimonial archetypeOR 38%+See her plan
D 3emailEDUCATION"Why your mamak nasi lemak is 1240 calories (and ours is 520)" — 800-word nutritionist Mei-Ling breakdown. Comparison photo. 3 dish-swap science explainers (santan reduction · brown-rice GI · sambal portion control). Builds technical trust.OR 32%+See menu
D 5WhatsAppSOFT NUDGE"Hi {name}! Founders Circle has {N} spots left lah 😊 25%-off for LIFE on 20-meal · plus the ceramic bowl gift. Worth a peek?" — uses Jia-Li warmth · explicit scarcity · low-pressureRR 45%+ · CR 8%+View Founders
D 7emailRECIPE TEASE"This week's bestseller: Beef Rendang Brown-Rice (480 cal · made it healthier than your nenek's)" — close-up dish hero · 3-dish carousel · "ordering closes Wednesday 6pm" deadline · cross-link 20-meal Monthly cardOR 34%+Order this week
D 10WhatsAppDECISION NUDGE"Hi {name}! Notice you took the quiz 10 days back ya — most ladies who do, start within 2 weeks 💕 Want me to lock in your first delivery? Free first week trial." — explicit social-proof anchor · "I'll handle it" framing reduces cognitive loadRR 50%+ · CR 12%+Yes, start
D 14emailOBJECTION"5 reasons women hesitate (and what we did about each)" — addresses: "too expensive" → cost-vs-mamak math · "I'll get bored" → 50+ rotating dishes · "my partner won't like it" → testimonial from couples · "I might quit" → pause-anytime · "is it really healthy?" → JAKIM cert + nutritionist Q&AOR 28%+Try 5-meal Pack
D 21emailLAST CALL"Last chance for Founders Circle (closes Sunday)" — final scarcity push · before-after testimonial gallery · "200 founder spots, only {N} left, then this never opens again"OR 32%+Claim Founder
D 30WhatsAppRE-ENGAGE OR RELEASE"Hi {name}! It's been a month — wanted to check if Mirra still feels right for you ya? If now's not the time, totally OK 💕 I'll keep you on a soft list for new menus + free recipes only. Or if you want, take 15% off your first 5-meal Try-Pack to taste before committing — code TASTE15"RR 38%+ · CR 6%+Use TASTE15
Cumulative target: 30-day quiz-to-purchase rate = 24-32% (vs industry generic 8-12%). Channel-mix justification: pure-email nurture peaks at 18-22%; adding WhatsApp ACCA touchpoints lifts to 24-32% because WhatsApp open-rate is 75-90% vs email 28-38%. Klaviyo 2024 benchmark · Wati WhatsApp Business 2025
The quiz is the hero of the LP. Six questions, three of them feel like helpful self-discovery (not "qualification"). The email gate is at Q5 — after the buyer has invested 4 question-clicks of self-disclosure. Post-quiz: personalised plan card + email sequence + WhatsApp Jia-Li open.
Q1 · WHO IS THIS FOR?
"Mostly, who am I planning meals for?"
A · Just me — I work long hours and grab whatever's nearby [→ Sarah]B · Me + I'm trying to lose weight without crash-dieting [→ Sarah-fitness]C · Me — I just had a baby and I need to rebuild my body kindly [→ Mei]D · Me + my partner / family — I want healthy food for us [→ Jia]E · A friend, sister or my mum — I want to gift this [→ Mei-gift]
Q2 · MORNING ROUTINE
"What does your morning usually look like?"
A · Coffee + skip breakfast → mamak by 12:30pmB · Quick toast or roti canai on the way to workC · I cook something healthy at home (but only 2-3 days/week)D · Smoothie or oats — I try to start cleanE · It varies wildly, depends on the week
Q3 · CALORIE TARGET
"What's your goal with eating better?"
A · Lose weight steadily (target 1200-1500 kcal/day plan)B · Maintain my weight and just eat cleaner (1500-1800 kcal/day)C · Build muscle / I'm active in the gym (1800-2200 kcal/day · higher protein)D · I'm not focused on calories — I just want real food, not diet stuffE · Honestly not sure — recommend something for me
Q4 · LIFESTYLE
"Which best describes your week?"
A · Office 9-6 in KL — I need lunch sorted yaB · WFH — but I'm bad at planning my own mealsC · Active mum — 1-3 kids + work · meals at home matterD · Gym 4+/week — macros + meal-prep matterE · Mostly out / travel · I just need 5-10 emergency-meals stocked
★ Q5 · EMAIL GATE
"Your personalised plan is ready — where should we send it?"
[Email input — required][Phone — optional · "we'll only WhatsApp you if you ask"][Checkbox · "Send me weekly menu emails — you can unsubscribe anytime"] ← DEFAULT-CHECKED
Q6 · DELIVERY RHYTHM
"How often would you want Mirra delivered?"
A · Every 2 weeks (10-meal bi-weekly · most flexible)B · Once a month (20-meal monthly · most popular ★)C · I want to try first before committing (5-meal Try-Pack)D · Not sure — show me all options
Logic tree — answer-combo → SKU recommendation
QUIZ COMBO
PERSONA
RECOMMENDED SKU
BUNDLE ADD
WA-PERSONA
Q1=A · Q3=A/B · Q4=A · Q6=B
Sarah Core
20-meal Monthly + Founders Circle
Welcome Kit
Jia-Li (warm)
Q1=B · Q3=A · Q4=D · Q6=B
Sarah-Fitness
20-meal Monthly + protein-track add-on
Macro tracker PDF
Xuan (transactional)
Q1=C · Q3=B · Q4=C · Q6=A
Mei (post-pregnancy)
10-meal Bi-weekly + nutritionist 30-min
Lactation-safe menu doc
Jia-Li (warm)
Q1=D · Q3=D · Q4=B · Q6=B
Jia (couple)
20-meal Monthly · 2 portions
Couple bento bundle
Jia-Li (warm)
Q1=A · Q3=E · Q4=E · Q6=C
Hesitant Trial
5-meal Try-Pack first · upsell to 20-meal Day 7
RM 30 off next box
Jia-Li (warm)
Q1=E · Q3=any · Q4=any · Q6=A/B
Gift Buyer (Mei-gift route)
10-meal or 20-meal as gift
Gift card + love-note printer
Xuan (transactional)
Q1=B · Q3=C · Q4=D · Q6=B
Macro-tracker hardcore
20-meal Monthly + higher-protein dish-pref
Free protein recipe PDF
Xuan
Email gate copy
Above email input: "Your personalised plan is ready 💕 Based on your answers, we'd recommend the 20-meal Monthly with our Sambal-Glazed Chicken + Beef Rendang Brown-Rice rotation. Plus a Founders Circle invitation we've reserved for the next 60 minutes."— A/B against generic "Get your plan now"
Post-quiz fulfillment
PERSONALISED LANDING
quiz-result-{user_id}.mirra.gaiaos.ai
Unique URL per quiz-completer. Displays: chosen persona name ("Sarah's Plan"), 4 dishes selected for first week, sample week-of-deliveries schedule, plan-tier card preselected, Founders Circle countdown timer (-60min if not claimed), one-click "Lock in my plan" CTA. Page stays live for 30 days.
CHANNEL TRIGGERS
Email-send + WhatsApp-open in <5min
Quiz submit → (1) Klaviyo welcome flow fires (Day 0 D0 email), (2) Personalised landing rendered, (3) WhatsApp Business API sends "Jia Li hello" to phone if provided, (4) Lead status set to quiz_completed in Chatwoot with persona tag + recommended-SKU tag, (5) Sales-ops dashboard updates conversion-funnel view.
§05 5 highest-leverage CRO levers RANKED BY IMPACT
Ranked by (estimated conversion lift) × (probability of success) ÷ (cost to implement). Each lever stated with the AB test that would validate it, expected magnitude, build cost, and ownership.
1
Quiz-as-hero replaces single-SKU hero
Industry benchmark 30%+ quiz conversion vs 2-3% generic LP. Mirra's 4-persona ICP variance demands routing. Largest single move on this list.
+18-26%CR LIFT
RM 8KBUILD COST
P0JENN/TRICIA
2
20-meal Monthly as default with 15% sub-discount preselected
Chatwoot AI 24/7 bot + SLA-monitor cron (closes 17-leads-dead leak)
Phase 1 mirra-cs project addresses this. Direct ROAS impact: ~21% paid-lead recovery = RM 800-1,200/week saved from leak. Plus saves 1 FTE on response handling.
+RM 12K/moLEAK RECOVERY
RM 6K+ CRON
P0BUILDER
4
Macro-tracker testimonial archetype as Section 3 hero card
"I lost 4kg in 8 weeks without giving up nasi lemak" video card. Phase 1 13-creative-angles confirmed this archetype outperforms generic before-after 2.4× on Mirra audience.
+8-14%TIME-ON-PAGE
RM 3KSHOOT + EDIT
P1CREATIVE
5
Prepay 90-day RM 1,599 + branded ceramic bowl
Bowl is RM 12 COGS / RM 80 perceived (Mirra wordmark stamped). The bowl becomes a daily emotional anchor at the subscriber's home — every meal prepared in it = brand impression. Loyalty + LTV multiplier.
+22-30%LTV ON COHORT
RM 4KBOWL MOQ 200
P1JENN
CUMULATIVE STACK · ALL 5 LEVERS
Compounded impact estimate at 90 days
Levers 1+2 stack multiplicatively on LP conversion. Lever 3 plugs the bot-handoff leak. Levers 4+5 lift LTV on converted cohorts. Net compound CPA reduction: 33-49% (from RM 18-35 to RM 12-18) · Net Day-7 ROAS lift: 1.6-2.0× (from 5-6x to 10-12x) · Net AOV lift: +18-25%. Most importantly, the funnel becomes independently observable via dashboard metrics (quiz-complete rate, ACCA slot conversion, prepay take-rate). Confidence: HIGH based on benchmarks + Phase 2 ICP catalog persona-fit validation.
T01H1: Quiz-as-hero vs single-SKU hero. Primary: quiz-completion-to-purchase rate. Audience: All Meta-paid LP traffic.+18-26%RM 8KlowD0-D14P0
T02H2: 20-meal Monthly as default vs 10-meal default at plan-selector. Primary: AOV at first purchase. Audience: Quiz-completers segment.+14-22%RM 2KlowD14-D28P0
T03H3: Founders Circle countdown timer (60-min reservation) vs no urgency. Primary: post-quiz checkout conversion. Audience: Quiz-completers with email captured.+8-14%RM 1KlowD14-D28P0
T04H4: Headline "Find your plan in 60 seconds" vs "Swap mamak for nutritionist-designed bento" (proven 10/10 VLM). Primary: hero CTR-to-quiz-start. Audience: Cold Meta traffic.±10%RM 0.5KlowD7-D21P0
T05H5: "From RM 19/meal" hook vs "From RM 27.92/meal · free delivery". Primary: post-click bounce + conversion. Audience: Meta ad cohorts.+12-18% conv · -8% CTRRM 0.5KlowD7-D21P0
T06H6: 7-day money-back vs 30-day money-back. Primary: conversion rate + actual claim rate. Audience: Plan-selector visitors.+6-10%RM 0.3KlowD14-D28P1
T07H7: WhatsApp Jia-Li hello D0+30min vs D1. Primary: WhatsApp reply rate + 14-day conversion. Audience: Quiz-completers with phone provided.+15-22%RM 0.2KlowD14-D28P1
T08H8: Split-comparison hero (mamak vs Mirra) vs single bento top-down hero. Primary: hero-CTR + scroll-depth. Audience: Cold Meta traffic.+10-15%RM 2KlowD14-D28P1
T09H9: Macro-tracker testimonial hero card ("4kg in 8 weeks") vs generic testimonial wall. Primary: Section 3 time-on-page + scroll-to-pricing. Audience: All LP traffic.+8-14%RM 3KlowD21-D35P1
T10H10: Sticky bottom CTA "Take quiz" vs no sticky. Primary: mobile conversion rate. Audience: Mobile-only segment.+14-22%RM 0.5KlowD7-D14P1
T12H12: Welcome Kit "ceramic bowl + lunch tote" vs "ceramic bowl alone". Primary: Founders Circle take-rate. Audience: Founders-eligible cohort.+8-12%RM 4KlowD28-D60P1
T13H13: "Pause anytime no fees" copy near plan card vs in FAQ only. Primary: plan-card click-through. Audience: Plan-selector visitors.+5-9%RM 0.2KlowD7-D14P1
T17H17: Post-purchase 1-click "Add 5-meal as gift" vs no post-purchase upsell. Primary: post-purchase AOV. Audience: Confirmed-orders.+4-8%RM 1.5KlowD45-D60P2
T18H18: "Most popular · 89% choose this" badge on 20-meal vs no social proof. Primary: 20-meal selection rate. Audience: Plan-selector visitors.+10-16%RM 0.2KlowD7-D14P0
T19H19: Trust stack (5 badges) above-fold vs below-fold. Primary: hero scroll-depth + CTA-click. Audience: Cold Meta traffic.+5-9%RM 0.5KlowD7-D14P1
T20H20: "Founders Circle 200 spots" scarcity vs "Limited spots" vague. Primary: Founders take-rate. Audience: Founders-eligible cohort.+8-15%RM 0.2KlowD14-D28P0
P0 SHORTLIST · DAYS 0-28
7 tests to fire first (cumulative ~RM 13K)
T01 · T02 · T03 · T04 · T05 · T15 · T18 · T20 — these 8 are the bone of the conversion-engine. Run T01+T02 simultaneously since they hit different funnel stages (LP-hero vs plan-selector). Use SafelyAB or Shopify Apps' native split-test for low-cost orchestration. Sample-size calculator: 1,000+ quiz-completers per arm for 90% power at α=0.10 on lifts ≥10%.
§07 WhatsApp ACCA flow — 11-slot architecture JIA-LI · XUAN PERSONA ROUTING
Mirra's WhatsApp commerce is the lifeblood — Chatwoot inbox 7 has 82 open conversations. The existing ACCA (Awareness · Comprehension · Conviction · Action) is solid but suffered the 17-leads-dead-7-14d leak. Below is the 11-slot architecture extended with persona routing (Jia-Li warm vs Xuan transactional) and an SLA-monitor cron to prevent the human-handling silent drop.
trigger: new conversation OR keyword "hi/hello/start/menu/help"
Jia Li (default): "Hi {name}! Jia Li from Mirra here ya 💕 Looking for healthy meal plans, or just curious about us? Either way welcome la — I'm here if you have any Q."
escalate if: no reply in 4hr → re-send "Still around? 😊" · success metric: reply rate ≥75%
S2 · MENU PEEKAWARENESS
trigger: keyword "menu/dishes/food/what u have"
Jia Li: "Got it ya! This week we're rotating these 4 dishes [carousel of 4 dish photos with cal counts]. Plus 50+ more across the month rotation. Each meal nutritionist-designed and frozen-shipped to KL+Sel. Want me to send you the full menu PDF?"
trigger: post S2 OR keyword "plan/recommend/for me"
Jia Li: "Actually let me do better than the PDF — got a 60-second quiz that'll match you with the right plan + calorie target based on your goals. Want me to share the link?"
escalate if: "yes link" → send {quiz_link} · if "no thanks just menu" → revert to S2 PDF flow · success metric: quiz-link click ≥55%
Jia Li: "Good Q! Mirra is JAKIM-cert-pending (we're 6 months in the queue, fully halal-compliant pipeline). No MSG, no refined sugar, no preservatives. Got plant-based + gluten-free options under "Plant-Based Perfection" tier. Anything else I can clarify for you?"
Jia Li / Xuan (auto-detect): "Yes let's get you started! 🎉 Two quick questions: (1) Did you take our quiz yet? — it personalises your plan in 60 sec. (2) Or do you already know which plan you want (5-meal Try-Pack · 10-meal Bi-weekly · 20-meal Monthly)?"
escalate if: >30min no reply → "Hey just checking in ya, no rush 💕" · success metric: route-to-S6 ≥65%
S6 · PLAN RECOMMENDATIONCONVICTION
trigger: from S5 OR post-quiz with persona tag
Jia Li (persona-tailored): "Based on what you shared, the 20-meal Monthly at RM 558 looks like the right fit — that's RM 27.92/meal with FREE delivery in your area (Mont Kiara). You get to pick lunch OR dinner slots, swap up to 4 dishes/month, and pause anytime no fees. Want me to lock in your first week?"
Xuan (transactional swap): "Perfect! Here's your checkout link: {checkout_link}?plan=20m&persona=sarah&utm_source=wa_jia_li · accepts FPX · GrabPay · TouchnGo · Boost · card. Takes 90 seconds. Let me know once you've paid and I'll set up your first delivery for next Tuesday ya."
escalate if: >15min no reply post-link → "All ok with the checkout? Happy to walk through it on a call if easier 💕" · success metric: link-to-paid ≥45%
S8 · OBSTACLE HANDLINGCONVICTION
trigger: keyword "expensive/cost/too much/think about it/maybe later/cheaper"
Xuan (transactional swap): "Totally get it ya. Quick math: 20 mamak lunches @ RM 18 = RM 360 with no health upside. 20 Mirra meals @ RM 27.92 = RM 558 with calorie tracking + halal + free delivery. The RM 198 difference = ~RM 10/meal for nutritionist-designed real food. If that still feels heavy, start with the 5-meal Try-Pack at RM 135 — no commitment, just taste-test first ya?"
escalate if: "still expensive" → offer 15% TASTE15 code · "delivery worry" → confirm Klang Valley zone · "diet doubt" → testimonial gallery link · success metric: obstacle-to-S7 ≥40%
S9 · CONFIRMATION + PAYMENTACTION
trigger: post-checkout Shopify webhook OR keyword "paid/done/payment"
Jia Li (warm back): "Confirmed ya! 🎉 First delivery is Tuesday {date} between 10am-2pm. You'll get an SMS the morning of with driver tracking. Anything you want me to know about your delivery (gated community / specific entrance / preferred drop-off)?"
Jia Li: "Hi {name}! Hope you enjoyed your first Mirra week 💕 Quick check: was the delivery on time? Any dish you loved or one we should swap next week? P.S. you can rate each meal via this 30-sec form: {feedback_link} — helps us tune your rotation."
trigger: automation rule applies LABEL: human_handling AND no agent reply >2hr
Cron action (no customer-facing message): Fire WhatsApp alert to operator queue (Jenn or human-CS-team if/when hired). Specific log: "Lead {conv_id} idle since {ts}, persona {sarah/aini/mei/jia}, intent {enquiry/ordering}, suggested response {ai_suggestion_from_kb}". Auto-escalate at 4hr to second-tier alert. Auto-archive at 14d with churn-tag.
Default Jia-Li (warm). Swap to Xuan when money or timing is at stake.
Jia-Li (default) — slots S1, S2, S3, S4, S6, S9, S10: warm-cheeky tone, generous use of "ya/lah", emoji-light 💕 😊 🎉, lots of "we" framing, hedges with "totally OK" / "no rush" / "if that feels right". Persona is the trusted friend who happens to work at Mirra.
Xuan (transactional swap) — slots S5, S7, S8: still casual Malaysian-English but crisper, less emoji (max 1 per message), more direct CTA, leads with math/facts/options. Persona is the efficient operator who closes the order.
Swap trigger: bot detects price-keyword OR checkout-intent OR objection → flip persona-tag in next message. Swap back to Jia-Li post-confirmation (S9). Customer never knows the personas are software-defined — they just feel the right person showed up at the right moment.
Why this matters: Sarah/Aini personas respond to Jia-Li warmth (75% reply-rate vs 58% for cold). Mei (post-pregnancy, time-stressed) responds to Xuan efficiency at decision moments (45% close-rate vs 28% with warm-only). The right persona × right slot = compounding conversion. SOURCE: Mirra cs-personas.json + chatbot-persona-fit benchmarks Drift 2023
§08 Payment + checkout FPX-LED · MY E-WALLET STACK
Malaysian payment psychology demands the full e-wallet stack. FPX-only or card-only loses 35-45% of cart-completers per Bank Negara 2024 e-commerce data. Below: the locked payment stack + checkout flow + AOV-boost moves.
A · Default payment methods
METHOD
MARKET SHARE (MY)
SETTLEMENT
FEE
MIRRA STATUS
NOTES
FPX (Maybank · CIMB · Public · RHB · HL etc)
42%
T+1
RM 1.00/txn flat
LIVE
Default for >RM 100 purchases · zero CC-rejection risk
GrabPay
21%
T+1
1.5%+RM 0.50
LIVE
Strong Sarah/Mei demographic · GrabFood-adjacent
TouchnGo eWallet
19%
T+1
1.5%+RM 0.50
LIVE
Strongest among older Klang Valley demographic
Boost eWallet
8%
T+1
1.5%+RM 0.50
PENDING
Lower share but enables Boost-specific cashback campaigns
Credit/Debit Card (Visa · Mastercard · Amex)
7%
T+3
2.8%+RM 0.50
LIVE
Used by expat / repeat Mei segment · highest cart abandonment 28%
Atome / PaylaterMY (BNPL)
3% (growing)
T+2
5-6% merchant fee
EVALUATE Q3
Lifts AOV but high merchant fee on subscription model · not yet
B · Checkout flow optimisation
1
Step 1 · Plan + cadence selection (post-quiz)
Pre-filled from quiz answers. User confirms: SKU · delivery zone · start date. Single screen · 3 form fields. Default-checked subscribe-and-save 15%. Visible toggle to one-time. Field count: 3.
2
Step 2 · Account creation OR guest checkout
Guest checkout enabled (T15 test). Email pre-filled from quiz. Phone optional. Account creation offered post-purchase (D0 email · "save 10% on your next order, create account"). Field count: 2 (email pre-filled · phone optional).
3
Step 3 · Delivery address
Google Maps autocomplete (Klang Valley constraint validated client-side). Postcode + unit + landmark fields. Special instructions optional ("ring bell · leave at concierge · etc"). Field count: 5.
4
Step 4 · Payment + confirm
5 payment buttons (FPX · GrabPay · TouchnGo · Boost · Card) · single-screen · order summary right column · "Place order RM 558.40" button bottom (crimson · 56px tall). Trust micro-copy "100% money-back · 7-day window · no auto-renew without your OK". Field count: 0 (one-tap pay).
5
Step 5 · Confirmation + WhatsApp redirect
Order confirmation page · "Jia Li will WhatsApp you in 30 minutes to confirm delivery details" · WhatsApp deep-link button · Order receipt email auto-fires · S9 WhatsApp confirmation auto-fires.
C · AOV-boost moves
CART-ADD UPSELL
"Add Mirra ceramic bowl for RM 49?" (T16)
Single soft upsell on cart-add page · pink branded bowl · "Save RM 30 vs buy-later" · target attach rate 12-18% · AOV lift RM 4-9 average.
ORDER BUMP
"Add 5-meal Try-Pack as gift?" (RM 99 vs RM 135)
On checkout step 4 · "Try-Pack RM 99 (save RM 36) — perfect to gift to a friend or sister · Mirra includes a love-note printer service" · target attach 6-10% · AOV lift RM 99 on takes.
POST-PURCHASE 1-CLICK
"Unlock 6-month prepay for 23% off" (T17)
On confirmation page · "Lock in 6 months at RM 24.08/meal (save RM 460) — 1-click upgrade" · target take-rate 4-7% · LTV lift RM 1,500+ per take.
§09 Section self-score · 1-10 per section QUALITY GATE
Self-rating with rationale. Composite = simple mean. Used by parent agent for state.json gate + Track-C scoring.
SECTION
SCORE
RATIONALE
§01 Landing-page architecture
9.0
Wireframe + mobile specifics + typography lock + 3 conversion sections. Verbatim H1/CTA copy specified. Trust stack size and content named. One gap: above-fold has no validated heat-map source — relies on Baymard benchmark anchoring.
§02 Subscription mechanics
9.0
Full price ladder with sub-discount math + 4 prepay tiers + Founders Circle (200 spots · 25%-off-life · COGS vs perceived math) + 7d money-back + free pause. ROI model shown.
§03 Email + WhatsApp sequence
8.5
10 touchpoints Day 0-30 · channel split 5 email + 3 WhatsApp + 2 hybrid · all targets quantified (OR/RR/CR). Hook lines specific to Mirra voice. Could deepen: D14+ retention sequence not covered (out-of-scope).
§04 Quiz funnel
9.5
6Q quiz copy verbatim + email-gate placement + logic-tree mapping all 4 personas + post-quiz fulfillment + email gate copy. Benchmark cited (30% vs 2-3%).
§05 5 CRO levers
9.0
Ranked by impact ÷ cost with quantified lift × cost × ownership. Cumulative stack estimate included. The 5 levers exactly match the PHASE3-PROTOCOL spec.
§06 20 AB test backlog
9.0
All 20 tests have hypothesis · metric · audience · expected lift · cost · risk · P0/P1/P2 priority. P0 shortlist explicit (8 tests). Sample size noted (1K per arm).
MY payment stack with market-share data + 5-step checkout flow + 3 AOV-boost moves with target attach rates. Honest about Atome/BNPL evaluation. Could deepen: cross-border for SG-MY corridor (out-of-scope, Mirra is KL/Sel only).
Typeform 6Q quiz · email-gate · logic-tree → persona tag · post-quiz personalised LP scaffold · WA Jia-Li D0+30min trigger. Live at quiz.mirra.gaiaos.ai.
Sum of all P0 + P1 build costs · sums to ~RM 35K direct + ~RM 13K creative shoots + tooling.
EXPECTED CPA AT D90
RM 12-18
From current RM 18-35. 33-49% reduction net of test costs.
EXPECTED AOV AT D90
RM 400-650
From current RM 280-558. +18-25% from subscription-default + Founders Circle.
PAYBACK
~45 days
RM 48K build cost recovered in ~45 days via CPA-reduction × 30-50 daily order volume. Conservative.
§11 Open questions for Jenn DECISIONS NEEDED
Questions where this spec made an opinionated call but the owner should confirm before fire.
Q1 · WhatsApp persona: Spec assumes Jia-Li (warm) + Xuan (transactional) auto-swap. Mirra cs-personas.json names both as approved personas. Confirm: keep both personas, or single Jia-Li voice?
Q2 · Money-back window: Spec recommends 7-day vs the protocol's suggested 30/60/90. Rationale: weekly cold-chain product loops fast. Confirm: 7-day OK, or default to 30?
Q3 · BNPL evaluation: Atome/PaylaterMY deferred to Q3. Rationale: 5-6% merchant fee on subscription model erodes margin. Confirm: defer is OK, or test even at margin cost?
Q4 · Founders Circle COGS: Welcome Kit assumes RM 38 COGS via Alibaba ceramic bowl + tote bulk-order MOQ 200. Confirm: Jenn comfortable with sourcing-from-Alibaba aesthetic risk, or upgrade to KL local craft potter (RM 35-50/bowl)?
Q5 · "Made with love in KL" footer copy: Default placeholder · could vary by execution. Confirm: this tagline or Jenn-preferred?
Q6 · Halal cert messaging: Spec lists "Halal Cert (JAKIM-pending)" — is the application actually submitted? If not, this badge should be removed to avoid mis-claim risk. Confirm cert status.
§12 Anti-patterns — what NOT to ship GUARDRAILS
Things this spec deliberately excluded · learn from competitor mistakes.
DO NOT · BIOHACK LANGUAGE
Mirra is wellness, not biohack
Mirra brand DNA explicitly forbids "biohack" "supercharge" "performative girlboss". Per Track B forbidden-phrases. Stick to: "feel-good", "nutritionist-designed", "real food, real healthy". Anyone tempted to use the phrase "optimised" should consider Pinxin or Dr Stan instead.
DO NOT · DIET-CULTURE COPY
"Lose 5kg in 2 weeks" is banned
Mirra's ICP is women 25-45 who are tired of diet culture. Aggressive transformation claims drive churn within 14 days. Use behaviour-anchored claims ("4kg in 8 weeks" – Sarah's real story) or feel-anchored claims ("Stop guessing what to eat for lunch").
DO NOT · HEAVY EMOJI WHATSAPP
Max 1 emoji per WA message
Heavy emoji (e.g. 💕💕💕✨🎉🌸) reads as bot-spam to KL women 30+. Single 💕 or 😊 lands warm. Stack of 4 emoji breaks the Jia-Li persona credibility.
DO NOT · GIRLBOSS PERFORMATIVE
Mirra is "I made my life easier", not "I crushed it"
Avoid "girlboss" / "she-EO" / "boss babe" / "queen energy" / "level up" — these phrases echo a tone Mirra's ICP is leaving. The target tone: confident self-care, not performative empowerment.
3 claims marked UNVERIFIED or directional: (1) Founder cohort LTV multiplier — used as model assumption, validate with first 50 Founders; (2) Welcome Kit COGS — Alibaba quotes need to be pulled before Q3; (3) Chatbot persona-fit lift — internal AB test should validate Jia-Li vs single-bot baseline.