Executive Summary

Floward’s data science team wanted to move beyond opaque, black-box A/B testing. In late 2024, they migrated to GrowthBook Enterprise, integrating directly with their AWS Redshift warehouse to bring experimentation inside their own data pipeline.

Within nine months, the team launched 200+ live experiments across web, iOS, and Android, validated results in real time, and helped drive double-digit revenue growth. More importantly, GrowthBook gave Floward complete visibility, control, and reproducibility, turning experimentation into a data-driven discipline and encouraging an experimentation culture within the company.

Background

Floward operates across nine  markets, including Kuwait, Saudi Arabia, Oman, Qatar, Bahrain, United Arab Emirates, Jordan, Egypt, and the United Kingdom. Each market behaves differently, from purchasing power to cultural preferences, so the data team needed to test locally, learn globally, and maintain a single source of truth.

The Challenge: Black-Box Experimentation

Before GrowthBook, experiments ran through a platform with limited custom metric support. That approach hit predictable pain points:

• No access to underlying metric definitions or SQL
  
• Difficulty establishing proper randomization for experiments
  
• Inability to augment the measurement framework to add or enrich insights
  

The Solution: Warehouse-Native Experiments

Floward adopted GrowthBook Enterprise to align experimentation with its existing AWS infrastructure.

Integration Overview

• Warehouse-native: All metrics computed directly from Redshift; no duplicated events.
  
• Custom SQL metrics: Engineers define standard conversion, revenue, and retention queries in SQL, ensuring consistency across experiments.
  
• Instant rollouts: Flags and variations can be turned on or off in seconds without requiring new builds.
  
• Granular targeting: Experiments segmented by country, language, and device type.
  

Product and Data Wins

Before GrowthBook, the setup relied on third-party aggregation, manual experiment configuration, and limited rollout control with basically zero ability to explain what was actually happening. After GrowthBook, everything runs on Redshift as a single source of truth, experiments are SDK-driven and versioned in Git, rollouts and rollbacks are instant, and results are fully explainable with transparent SQL and daily monitoring.

“GrowthBook lets us build experiments exactly how we want. The ability to target based on the culture and geography as granular as we want is a major hit for us.”
— Eslam Samy, Data Scientist

Experiments That Moved the Needle

Localized Homepages

Different homepage layouts and collection priorities for Saudi Arabia and Kuwait improved cultural relevance and boosted homepage-to-checkout conversion.

• Lifted conversion rates 
• Validated in < 2 weeks with full statistical significance

Post-Order Experience

Optimized recipient delivery flow (slot confirmation, timing, feedback). Result: measurable retention gains and higher customer satisfaction.

Delivery Fees & Free-Delivery Thresholds

A/B tests found market-specific free-delivery “sweet spots” that maximized order value without hurting margins.

Quantitative Impact

• Experiments Run: 200+ across three platforms‍
• Experiment Velocity: Setup time cut from 3 days to less than 30 minutes‍
• Sales Growth: Double-digit YoY increase‍
• Decision Latency: Daily monitoring vs weekly reports‍
• Cultural Adoption: Product & commercial teams self-serve reporting

Key Learnings for Engineering Teams

• Own your data path. Run metrics from your warehouse, not your vendor’s.
  
• Start small, scale fast. Prove value with one or two experiments, then templatize rollouts.
  
• Enable every role. Provide product managers with access so data scientists aren’t bottlenecks.
• ‍Collaborative monitoring. Daily checks mitigate experiments not going in the right direction.

‍