Supply Chain Analytics

Built 97% accurate sales forecasts across SKUs for FMCG company to optimize inventory costs

Request a FREE Proposal
Built 97% accurate sales forecasts across SKUs for FMCG company to optimize inventory costs
Business Outcomes
brand level accuracy achieved
location level accuracy achieved
Reduced inventory costs and maximized sales productivity
Reduced inventory costs and maximized sales productivity

Known for its high-quality juices, beverages, dairy products and snacks, the client is the Nigerian division of a leading US-based FMCG giant in the Food and Beverages industry. The company sells all its products through thousands of depots spread across the African nation.

Business Need

The traditional sales forecast methods used by the beverage company to predict sales across geographies and product lines was replete with errors and inaccuracies.

The primary problem areas with the legacy forecast methodology were:

  • Low visibility into data and key information across business functions and systems
  • Inaccurate and obsolete data on critical parameters were feeding into forecast calculations
  • Inconsistent data availability on market promotions, discounts etc. flawed demand calculations

The company partnered with Hitech to develop a robust sales forecasting methodology to:

  • Accurately forecast daily sales and need for replenishment of perishable products
  • Streamline end-to-end inventory to reduce storage costs and sales bottlenecks

Want to build a connected, intelligent, scalable and sustainable supply chain?

Get in Touch with us »


The team at Hitech Analytics delivered a smart sales forecasting solution to accurately predict sales of each product across product lines and locations. The machine learning driven statistical model leveraged time series forecasting, regression and boosting algorithms to enhance accuracy of forecasts.


  • We received annual data of last three years on regions, sales, categories, types of products, costs, etc. from the beverage company. Our data scientists cleansed, standardized and segregated relevant data from voluminous and inconsistent data sets for more than 1280 SKUs across locations and siloed systems.
  • During initial assessment, the team found that the accuracy of legacy systems stood at 18.4% for location and 12.6% for brands. To select the best fit model, we combined various machine learning algorithms like XG-Boost, Bayesian, SARIMA and ARIMA.
  • Our data experts developed sales forecasting solution using different set of models with customized parameters for right estimation of product replenishment. It also factored in the influence of seasonality, holidays and economic trends and aligned with the surge in customer activities after promotions and discounts.

Software & Technology Used

Python, Excel and ML-algorithms


The consumer goods company received granular and accurate 6-monthly forecasts through interactive and intuitive dashboards. With accuracy in forecasts and better inventory allocation, they were able to increase sales, maintain stocks of in-demand products, minimize wastage of low-demand SKUs and improve reach among customers.

Let's get started.

Talk to us to build fresh perspectives and leverage smart innovation for winning business outcomes.

Get in Touch
Contact Us backtotop