Coca-Cola uses AI to improve retailer ordering in Malaysia
Coca-Cola deployed "Perfect Basket," an AI-powered recommendation feature within its Coke Buddy B2B platform, to suggest optimal product assortments and quantities to Malaysian retailers The Central Recommendation Engine analyzes multiple data signals including previous orders, ordering frequency, seasonality, weather patterns, and purchasing trends among comparable businesses During a January–April 2026 campaign, 83% of the 4,000+ participating retailers adopted Perfect Basket recommendations,
Analysis
TL;DR
- Coca-Cola deployed "Perfect Basket," an AI-powered recommendation feature within its Coke Buddy B2B platform, to suggest optimal product assortments and quantities to Malaysian retailers
- The Central Recommendation Engine analyzes multiple data signals including previous orders, ordering frequency, seasonality, weather patterns, and purchasing trends among comparable businesses
- During a January–April 2026 campaign, 83% of the 4,000+ participating retailers adopted Perfect Basket recommendations, with those outlets reporting higher sales revenue growth compared to non-participating peers
- Coca-Cola has scaled similar AI-enabled suggested-order capabilities to over 3 million outlets in Latin America and nearly 8 million B2B customers globally, indicating a broader strategic push into AI-driven supply chain optimization
- The company reports that AI-generated recommendations increased retailer purchase likelihood of recommended SKUs by over 30% in earlier pilots, and a three-country demand-prediction pilot achieved 7–8% higher sales versus non-AI outlets
Why It Matters
Coca-Cola's Perfect Basket demonstrates how large CPG companies are operationalizing AI to transform traditional B2B ordering workflows, shifting from reactive sales-driven replenishment to predictive, data-driven recommendations. For AI practitioners, this case illustrates the practical integration of multi-source data (historical, environmental, and peer-based) into production recommendation systems that directly impact revenue and inventory efficiency. The initiative also highlights the strategic value of AI in reallocating human sales resources toward higher-value account development rather than routine order-taking.
Technical Details
- Central Recommendation Engine: The core AI system behind Perfect Basket, aggregating and analyzing heterogeneous data signals—previous order history, ordering frequency, seasonality, weather data, and cross-retailer purchasing patterns—to generate product-and-quantity recommendations
- Multi-channel deployment: Coke Buddy supports ordering via mobile app, website, and WhatsApp, with Perfect Basket recommendations surfaced before order submission, allowing retailers to review and modify suggestions before finalizing purchases
- Data architecture: The system combines internal customer transaction data with external signals (weather, geolocation) and peer benchmarking data from comparable retail outlets to produce context-aware replenishment recommendations
- Global scale: AI-enabled suggested-order capabilities have reached over 3 million outlets in Latin America, with nearly 8 million B2B customers connected across Coca-Cola's global bottling network
- Performance metrics from pilots: Earlier pilots showed a 30%+ increase in retailer purchase likelihood for recommended SKUs; a three-country demand-prediction project combining historical sales with weather and geolocation data achieved 7–8% sales uplift versus control outlets
Industry Insight
- AI-driven B2B commerce is maturing beyond novelty: Coca-Cola's deployment across tens of thousands of outlets with measurable revenue impact signals that AI recommendation engines for wholesale and distribution are transitioning from experimental pilots to core operational infrastructure in CPG and adjacent industries
- Hybrid human-AI sales models are the emerging standard: Coca-Cola explicitly retains sales representatives alongside the digital platform, suggesting that the most effective enterprise AI implementations augment rather than replace human relationships—particularly in sectors where trust and account management drive long-term value
- Multi-signal recommendation engines deliver outsized ROI: The combination of internal transactional data with external contextual signals (weather, seasonality, peer behavior) appears to be a key differentiator in recommendation accuracy, suggesting that organizations should prioritize data integration and feature engineering over model complexity when building production recommendation systems
Disclaimer: The above content is generated by AI and is for reference only.