The Work/Cutting Inventory Costs by 96%
Toffee Inc.

Cutting Inventory Costs by 96%

A chocolate bar, five states, and a cost problem hiding in a spreadsheet.

Research
Overview
Modelled result — not independently verified

A confectionery company's demand kept swinging wildly by season. A demand-forecasting model points to a modelled 96% reduction in inventory costs.

Outcome

Total inventory costs modelled down from ₹31,25,060 to roughly ₹1,28,749 — a 96% reduction, based on five years of historical data, not an independently verified real-world result.

Why was a chocolate bar's inventory bleeding cash?
  • Huge seasonal demand swings, a 15-day lead time on supply, high ordering costs, and constant pressure between keeping enough stock and not overstocking a product that spoils.
Could AI have done this better?
  • AI wasn't used in the original model — the 96% cost reduction came from a seasonally-split statistical forecast, not a machine-learning system.
  • The natural next step is an ML-based demand model trained on the same five years of sales data, to catch finer seasonal and regional patterns than a split annual/peak model can — a recommendation, not something built here.
What's the actual fix?
  • Forecast demand by season rather than with one annual model, then calculate the ideal order size separately for peak and off-season.
  • Paired the forecasting fix with vendor coordination and cold-chain logistics recommendations, addressing the spoilage risk directly.
What's the model built on?
5 years of sales & cost dataSeasonal demand forecastingOrder-size optimisation
What's the proof?
  • Total inventory costs modelled down from ₹31,25,060 to roughly ₹1,28,749 — a 96% reduction — plus five practical recommendations.
  • A modelled result from historical data, not an audited live deployment.
Reflection
"A fairly simple model can produce a dramatic result, if you split it by the one variable that actually matters."
Project Report

Read the complete report submitted for this project, including the research, analysis, strategy, and final recommendations.