Building a Growth Engine From Zero in the World's Most Price-Sensitive Market

2. Context

Company: Shikho — Bangladesh's leading edtech startup. B2C subscription app for SSC and HSC exam preparation (high school students, grades 6–12). Also operates Bohubrihi, a professional skills platform. Named to Forbes Asia 100 to Watch (2022). Total funding: $8M.

My role: Lead, Performance & Growth Marketing (Apr 2024 – present). Full ownership of paid acquisition, growth marketing, lifecycle marketing, measurement infrastructure, and lead generation for both Shikho App and Bohubrihi. Built and led a 4-person growth team.

Market reality: Bangladesh. 170 million people. Android-dominant. Average revenue per user $3–$12/month. Parents are the payers; students are the users. bKash and Nagad mobile wallets are the primary payment methods — most users have no credit card. Facebook Messenger, not WhatsApp, is the dominant messaging platform. $35K/month paid media budget.

3. Challenge

When I joined, Shikho had strong product-market fit but no growth infrastructure. No mobile measurement partner. No attribution beyond platform-reported numbers. No lifecycle marketing. No experimentation framework. No cohort-level analysis. The company was spending on paid acquisition without any way to measure true impact or optimize beyond surface-level metrics.

The deeper challenge: how do you build a sophisticated growth engine — the kind of measurement and experimentation infrastructure that companies like Agoda and Duolingo operate — with 4 people and $35K/month, in a market where the tools, partners, and talent pool that exist in Singapore or Bangkok simply don't?

The constraint became the thesis: if you can build it here, you can build it anywhere.

4. Approach & Strategy

Layer 1 — Measurement First (Months 1–6)

Before scaling anything, I implemented the measurement stack. Led the company's first MMP implementation (Adjust), designing the event taxonomy from scratch — 15 key in-app events mapped to acquisition → activation → revenue. Built BigQuery pipelines connecting Firebase and GA4 data. Created Looker Studio dashboards for the CEO, the paid team, and the lifecycle team. Implemented server-side bKash/Nagad payment event tracking via custom webhooks — non-trivial in Bangladesh, and a critical proof-point for any market with non-standard payment infrastructure.

Layer 2 — Experimentation Culture (Months 3–12)

Formalized our experimentation framework: every test gets a hypothesis, baseline metric, minimum detectable effect, decision rule, and documented learning. Set up a structured experiment log that grew to 700+ documented experiments over 24 months. Maintained statistical rigor: pre-calculated sample sizes, 95% confidence thresholds, pre-registered primary metrics. Installed GrowthBook for feature-flag-based A/B testing, reducing dependency on engineering sprints.

Layer 3 — Full-Funnel Optimization (Months 4–18)

Implemented CleverTap for lifecycle marketing, building 5-state user segmentation (New → Current → At-Risk → Dormant → Resurrected) with automated flows for each transition. Designed a creative testing framework producing 30+ ad variants per month with systematic hook-message-format isolation. Launched TikTok as a new acquisition channel from zero. Ran ASO optimization in Bangla. Built a referral program. Created parent-specific WhatsApp lifecycle programs. Optimized the paywall through 15+ A/B tests on placement, trial length, and pricing structure.

Layer 4 — Causal Measurement (Months 7–24)

Designed and ran 4 geo-lift incrementality studies — suppressing paid in control regions to measure true incremental impact. Built a lightweight media mix model to inform quarterly budget allocation. Created an incrementality-calibrated attribution model that adjusted Adjust data with geo-lift coefficients. This is the layer that separates a good growth marketer from one that companies like Noom and Booking.com would fight to hire.

5. Results

Metric	Before	After	Change	Source
Install-to-conversion rate	1.73%	5.2%	3x improvement	Actual
Ad spend efficiency	Baseline	+53%	53% efficiency gain	Actual
MAU	Baseline	550K+	+36.7% growth	Actual
ARPU	Baseline	+34.8%	34.8% increase	Actual
Registration drop-off	Baseline	−27%	27% reduction	Actual
Monthly churn	Baseline	−29%	29% reduction	Actual
Active paid subscribers	Near zero	55K+	Built from scratch	Built
LTV:CAC ratio	<1.5x	4.2x	Sustainable growth	Built
Experiments documented	0	700+	Culture created	Built
Incrementality studies	0	4	Causal measurement	Built
India market launch	—	2.5K paid users	In 30 days	Actual

6. Team I Built

Built a 4-person team structured into two outcome-based pods: Pod 1 (Paid Performance & Measurement) and Pod 2 (Lifecycle & MarTech). Each pod with defined OKRs, weekly review cadences, and documented playbooks. Also managed relationships with 5+ external teams: Brand, Creative Production, Knowledge, Product, Engineering, and BI.

Created 3 operational playbooks (campaign launch, creative testing, attribution QA) that enabled the team to run independently — answering the hiring manager question: "what happens when you leave?"

7. Key Learnings

Retention improvement has 2.5x the impact of CPA reduction, dollar for dollar. Our sensitivity analysis proved this, and it fundamentally shifted our prioritization from acquisition efficiency toward activation and early retention.

The constraint was the advantage. A $35K/month budget forced measurement discipline that larger budgets never develop. When every dollar must prove its incremental value, you build systems that a $5M operation would benefit from but rarely bothers to create.

Measurement before scale, always. We discovered through incrementality testing that 22% of our attributed conversions were organic cannibalization. Without that test, we'd have continued wasting $4K/month — and never known.