Client
Mall91
Vision

Bring wholesale prices to Bharat shoppers via group deals.

Tech baseline

PHP Laravel monolith, MySQL, Android Kotlin client

Mall91

Mall91
Case Study

Executive Summary

Mall91 set out to reimagine e-commerce for India’s tier-2 and tier-3 consumers by combining local-language video selling, group discounts, and social referrals. A showcase Android MVP went viral on TikTok, yet growth exposed fatal cracks: checkout failures during flash sales, brittle warehouse integrations, and translation debt across ten languages. Investors demanded a production-grade, cloud-native platform—live before the Diwali shopping peak—or risk down-round valuation.

Steady Rabbit mobilised a Core-Flex Micro-GCC squad (Core engineers, Flex logistics & language-AI SMEs, Buffer bench). In ten two-week sprints we:

  • Re-architected the PHP monolith into event-driven Go micro-services on AWS EKS that scale to 4 M peak concurrent sessions
  • Introduced a rule-based warehouse orchestration layer, shrinking order-to-dispatch SLA from 26 hours to 6 hours
  • Implemented a linguistic-AI translation pipeline that auto-localises product feeds into 10 Indian languages with 93 % BLEU accuracy
  • Added a “drop cart” referral engine fuelling virality (K-factor 1.4) and CAC reduction of 37 %
  • Achieved 45 M registered users, ₹30 crore GMV/month, and p95 checkout latency of 620 ms during Diwali

Every sprint landed on time; the predictability premium paid back in under one festive cycle through cost-of-delay savings and GMV lift.

Client Profile & Business Context

  • Client
    Mall91

    Gurugram-based social-commerce start-up

  • Founded

    2018

  • Funding (Pre-Engagement)

    Seed + Series A (₹58 crore)

  • Vision

    “Bring wholesale prices to Bharat shoppers via group deals.”

  • Tech baseline

    PHP Laravel monolith, MySQL, Android Kotlin client

  • Strategic goal

    Hit 40 M users & ₹25 crore GMV before Diwali to unlock Series B

Problem Statement / Key Challenges

Flash-sale meltdowns

Challenge

Hard-coded AdMob only; ad CTR 1.6 % vs. market 2.0–2.2%.

Why It Threatened the Business

Lost revenue & user trust

Slow fulfilment

Challenge

Manual CSV exchange with 8 regional warehouses; SLA 26 h

Why It Threatened the Business

High cancel-rate, NPS 37

Translation debt

Challenge

10 languages managed by freelancers; 3-week lag.

Why It Threatened the Business

40 % catalog untranslated; bounce 31 %

Virality plateau

Challenge

Referral K-factor stuck at 0.9; CAC ₹342

Why It Threatened the Business

Unsustainable marketing spend

Investor deadline

Challenge

Diwali peak in 5 months; Series B contingent on scale KPIs

Why It Threatened the Business

Delay = down-round

Our Approach

Micro-GCC Squad Design

Layer
Roles
Mandate
Core (7)
Squad Lead/PO, 2 React-Native engineers, Go micro-services dev, DevOps/SRE, QA Auto, Product Analyst
Re-platform, release mobile super-app, achieve scale KPIs
Flex (2)
Logistics Integrations SME, NLP/Translation Engineer
High-risk spikes: warehouse APIs, NMT pipeline
Buffer (1)
Shadow Mobile Dev
Covers PTO/attrition—no cost to Mall91

Shift-Left Governance

  • Seven Plan-Left gates; SteadCAST flags risk/capacity daily.
  • Weekly 30-min steering with founders + ops lead.

Discovery Sprint 0

  • System-Thinking Workshop – Map flow: influencer video ➜ group cart ➜ payment ➜ warehouse wave ➜ delivery.
  • Architecture Blueprint – Go/gRPC services, Kafka event bus, Redis cart cache, warehouse adapters.
  • North-Star KPIs – p95 checkout ≤ 700 ms, warehouse ACK ≤ 2 h, translation lag ≤ 24 h, K-factor ≥ 1.2.

Outcome: Backlog 118 SP/sprint; launch locked for Week 20 (T-2 weeks before Diwali).

Solution Delivered

Cloud-Native Back-End

  • Go + gRPC services: Cart, Order, Payment, Referral, Translation..
  • Kafka for order & inventory events; Kafka-Connect sinks to Redshift.
  • Redis Cluster holds cart state; 99 th latency 2.4 ms.
  • p95 checkout latency 9 s → 620 ms.

Mobile Super-App

  • React-Native container (92 % shared code); mini-apps lazy-load via CodePush.
  • Integrated short-video feed (FFmpeg thumbnails < 1 s).
  • Crash-free sessions 99.83 % (Play-Store high-quality badge).

Warehouse Orchestration

  • Adapter pattern supports Shiprocket, Delhivery, Shadowfax.
  • Events trigger AWS Step Functions wave planning; SLA ACK 26 h → 6 h.
  • Fill-rate 96.4 % (↑11 pp).

Linguistic-AI Pipeline

  • Marian-NMT models fine-tuned on 0.6 M sentences; BLEU 93 %.
  • Auto-localises 250 k SKUs nightly; translation lag 3 weeks → 6 hours.

Viral Referral Engine

  • Drop Cart” feature: user needs 2 more buyers for big discount.
  • Real-time progress via WebSocket; K-factor 0.9 → 1.4; CAC ₹342 → ₹215.

Observability & Compliance

  • OpenTelemetry → Grafana Loki; SLOs: latency, error budgets, ACK SLA.
  • PCI DSS SAQ-A via Razorpay tokenisation.

Execution Journey

Sprint
Focus
KPI Shift
Predictability
Sprints 0
Discovery, arch
Baseline checkout 9 s
100 % gates
Sprints 1
EKS, Auth, Cart
p95 9 s→4.2 s
Risk WIP 17 %
Sprints 2
Kafka, Redis, video feed
Crash-free 97.1 %→99.1 %
Buffer unused
Sprints 3
Warehouse adapters PoC
ACK 26 h→11 h
Flex Logistics 24 h
Sprints 4
Translation NMT POC
Lag 3 wks→36 h
Flex NLP 24 h
Sprints 5
Referral MVP
K-factor 0.9→1.2
No slip
Sprints 6
Blue/green, load 1 M DAU
p95 4.2 s→700 ms
--
Sprints 7
NMT prod, leaderboards
Lag 36 h→6 h
Hot-fix 0
Sprints 8
e-wallet payouts, iOS launch
Crash-free 99.83 %
Budget +4 %
Sprints 9
Diwali readiness, audits
Users 30 M
Delivered 2 d early
Sprints 10
Post-launch tuning
GMV ₹22 Cr→₹30 Cr
—all green

Buffer dev covered React-Native engineer (viral fever) in Sprint 6—velocity dip 0 SP.

Business Outcomes & Impact

Registered users 8 M → 45 M inside 5 months

GMV ₹6 Cr → ₹30 Cr/month (5×) during Diwali

Checkout p95 latency 9 s → 620 ms (14× faster)

Order-to-dispatch SLA 26 h → 6 h (–76 %)

Translation lag 3 weeks → 6 hours; bounce rate 31 % → 18 %

Referral K-factor 0.9 → 1.4; CAC –37 %

Crash-free sessions 99.83 %; 4.4 Play-Store rating

Secured Series-B ₹130 crore at 3× valuation, citing “enterprise-grade scalability.”

Predictability premium (~7 % uplift) paid back in < 1 festive cycle by preventing a two-sprint slip worth ~₹6 Cr lost GMV.

Why Steady Rabbit?

Core-Flex Micro-GCC

Logistics & NLP SMEs inserted in 48 h; Buffer bench erased leave risk.

SteadCAST Predictability

98 % sprint adherence on 20-week roadmap.

Shift-Left Gates

Re-work –41 % with < 2 h overhead per sprint.

Domain Insight

Retail & social-commerce playbooks accelerated design decisions.

Outcome-Linked Engagement

KPIs (checkout latency, GMV, K-factor) tied to squad incentives.

Transparent Partnership

Weekly demos, Slack war-room, open burn charts—zero surprises.

Client Testimonial

Steady Rabbit

Co-Founder & CEO

Mall91

Steady Rabbit delivered the scale and speed Diwali demanded. Our shoppers saw flash-sale performance like never before, and investors loved the numbers. The Micro-GCC model is how you build certainty.