top of page
Bg.png

 WEBINAR ON DEMAND 

Modernize Your Analytics and Accelerate Your Move to Looker with Confidence

Migrating your Business Intelligence platform to Looker presents a powerful opportunity to modernize your data stack, but it requires careful planning and execution. This webinar provides a strategic roadmap for navigating the complexities of migration, from initial assessment to final rollout. We will focus on turning common challenges into strategic advantages by leveraging proven best practices and automation tools, ensuring you build a scalable and trusted analytics foundation on Looker.

Revolutionizing Inventory Management with AI for d.light

Updated: Jul 14

Efficient inventory management is essential for businesses aiming to balance operational efficiency and customer satisfaction. d.light, a global leader in solar-powered consumer products, sought to enhance its supply chain with AI-driven forecasting models. This initiative aimed to minimize costs while ensuring product availability for their dealer network worldwide.


image

Key Challenges


The customer needs to address the following inventory optimization challenges:


  1. Accurate daily forecasts: Accurate daily demand forecasts were needed to avoid stock-outs and overstocking, particularly for products with low or zero demand across some dealers.


  2. Inventory carrying cost reduction: Reducing inventory carrying costs and freeing up cash flow


  3. Dynamic product replacement: Ensure the model can adapt to frequent product retirements and new product introductions. The forecasting model should dynamically incorporate product replacements, allowing for continuous updates and adjustments based on new product availability.


The Solution


Partnering with SquareShift Technologies Inc., d.light implemented a robust solution leveraging Google Cloud Platform (GCP) and AI/ML technologies. The approach included:


  • Automated Model Management: Developing native forecasting models with automated refitting mechanisms to ensure optimal performance.


  • Advanced Pipelines: Utilizing Vertex AI pipelines for model development, forecasting, refitting, and tracking with the help of algorithms like LightGBM and Linear Regression.


  • Segmentation-Based Forecasting: Dealer segmentation by sales volume allowed tailored model development and dynamic routing for precise daily forecasting.


  • Dynamic Adaptation: Monthly evaluation of refitted models ensured consistently high-performing algorithms were applied.


The Impact


The project delivered significant results for d.light:


  • 15% Reduction in Inventory Costs: Optimized stock levels reduced overhead expenses.


  • 10% Decrease in Stock-Out Incidents: Enhanced demand forecasting ensured better product availability.


  • 12% WMAPE Accuracy: Precise forecasting supported improved inventory and procurement decisions.

 
 
 

Comments

bottom of page