Consumer Goods / FMCG
·Southeast Asia·11 monthsAI Demand Forecasting and Supply Chain Optimisation for a Consumer Goods Manufacturer
Deployed AI-driven demand forecasting and supply chain optimisation for a major consumer goods manufacturer distributing across Southeast Asian markets. The system generates SKU-level demand forecasts by market, optimises production scheduling against capacity constraints, and automates distributor replenishment planning to reduce both stockouts and excess inventory.
25%
Improvement in forecast accuracy
Thousands
SKUs forecasted by market by week
18%
Reduction in inventory holding costs
Challenge
Consumer goods companies managing thousands of SKUs across multiple markets face a demand planning challenge that compounds at every level. This manufacturer distributed hundreds of product lines through a network of distributors across Singapore, Malaysia, Indonesia, Thailand, Vietnam, and the Philippines. Each market had different consumer preferences, competitive dynamics, seasonal patterns, and distribution structures. Demand planning was performed by market-level planners using a combination of historical sales data, promotional calendars, and market intelligence — assembled in spreadsheets that grew more unwieldy with every planning cycle. Forecast accuracy directly impacted both revenue and cost. Overforecasting led to excess inventory — particularly painful for products with limited shelf life — that required markdowns or disposal. Underforecasting led to stockouts at distributor and retail level, driving consumers to competitor products. The bullwhip effect amplified forecast errors through the supply chain: a 10% demand signal distortion at retail could translate into 30-40% production planning variance by the time it reached the factory. Production scheduling was constrained by shared manufacturing lines, changeover times between products, raw material lead times, and capacity limits that varied by factory and line. The production planning team optimised within these constraints using legacy MRP systems that weren't designed for the demand variability and promotional intensity of modern FMCG markets. The disconnect between demand planning (performed by commercial teams in market) and production planning (performed by operations teams at factories) created a persistent gap that manifested as either excess inventory or missed demand.
Approach
Reluvate built an integrated demand-to-production planning platform. The demand forecasting module generates SKU-market-week level forecasts by combining statistical time series models (capturing seasonality, trends, and baseline demand) with ML models that incorporate causal factors: promotional plans, pricing changes, competitor activity, weather patterns, and macroeconomic indicators. Forecasts are generated at the granularity the business operates at — individual SKU by individual market by individual week — rather than the aggregated level that traditional planning tools support. The production scheduling module takes demand forecasts as input and generates optimised production plans that balance multiple constraints: line capacity, changeover efficiency (sequencing similar products to minimise changeover time), raw material availability, and minimum batch sizes. The optimisation explicitly accounts for uncertainty in the demand forecast — building appropriate safety stock for high-uncertainty SKUs while maintaining lean inventory for stable-demand products. Distributor replenishment planning automates the calculation of order quantities for each distributor based on their current inventory position, forecasted demand in their territory, and logistics constraints (order frequency, minimum order quantities, lead times). The system replaces the manual process where distributors placed orders based on their own estimates, which often bore little relation to actual demand patterns. Replenishment recommendations are shared with distributors through a web portal where they can review, adjust, and confirm orders.
Design Notes
The forecasting architecture was designed to handle the 'promotional planning' challenge that plagues FMCG forecasting. Promotions can increase demand by 200-500% for a promoted SKU, but the effect varies by promotion type, depth, timing, and market. Rather than treating promotions as a binary flag, Reluvate built promotion effect models that estimate the lift for each combination of promotion attributes and SKU characteristics. These models are calibrated on historical promotional performance and continuously updated as new promotions execute. Change management addressed the human factors in demand planning. Market planners had developed intuition about their markets over many years, and their judgment captured factors that data alone couldn't — upcoming competitor launches, retailer relationship dynamics, regulatory changes. Reluvate designed the system to present AI forecasts as a baseline that planners adjust with their market intelligence, rather than a final answer. The system tracks planner adjustments and measures whether they improve or degrade forecast accuracy, providing feedback that helps planners calibrate their own intuition. Exception handling covers the cascading failure modes in supply chains. When a raw material shortage constrains production capacity, the system re-optimises the production plan to prioritise the highest-value SKU-market combinations, recalculates distributor replenishment to reflect the constrained supply, and alerts commercial teams about potential stockouts so they can adjust promotional plans. This integrated exception handling prevents the siloed responses — commercial promoting a product that operations can't produce — that are common when planning functions operate independently.
Result
Demand forecast accuracy improved significantly at the SKU-market level, reducing both stockouts and excess inventory. Production scheduling efficiency increased through better demand signals and constraint-aware optimisation. Distributor replenishment shifted from reactive ordering to proactive, data-driven planning, reducing both overstock and stockout incidents at the distributor level. The bullwhip effect was measurably dampened as better demand signals flowed upstream to production planning.
Related Case Studies
Retail
Autonomous Operations for a National Furniture Retailer
Deployed end-to-end autonomous operations across five business functions — HR, procurement, customer service, accounts, and marketing — for a national furniture retailer with multiple showrooms. This is Reluvate's longest and deepest client engagement, running continuously for over 2.5 years with expanding scope.
Events & Hospitality
AI Customer Engagement Platform for a National Events Company
Deployed a multi-channel AI chatbot for a national events company, handling attendee inquiries, event registration, vendor coordination, and post-event follow-up across WhatsApp, web chat, and social media channels. The system reduced customer service headcount requirements during peak event periods while improving response times from hours to seconds.
Insurance
AI Claims Processing Automation for a Regional Insurer
Long-running engagement automating insurance claims processing for a regional insurer. The system handles document extraction from claim submissions, automated adjudication for standard claims, fraud signal detection, and workflow orchestration for complex claims requiring human review. Weekly collaborative meetings have run continuously for over two years.