The lack of understanding of economic activities, competitor review etc. makes it difficult to predict future sales accurately.
Client may overspend on marketing campaigns with minimal impact, overstock on unpopular items, or understock on in-demand furniture.
The company tend to hold excessive inventory of slow-selling items, leading to storage costs and lost sales opportunities for faster-selling furniture.
Solution.
Data Collection:Collect & integrated all the relevant data; e.g.
Sales, Inventory, Operating Cost, Margins, macro-economic indicators, seasonality, customer info etc.
Decision trees: Identify classification segments and key drivers;
e.g. Modeling at a county/zip level was determined to be most suitable for sales. Inventory management was determined to be performed at a state level
Ensemble Models & Neural Networks: Used to predict the sales revenue by considering all the drivers and their predictions into single forecast.
ARIMA & Prophet: Leveraged to capture the autocorrelation and seasonality patterns in products, and generate forecasts based on historical trend
e.g. Sales surge occurs in May-August period due to good weather
Optimization Modeling: What-if scenarios for the given pricing, marketing spend, and staffing levels that will maximize sales revenue
The Impact.
Improved Sales forecasting accuracy up till 91%,
Ensuring that popular products are always in stock with 95% confidence
More informed decision-making, as the seller can understand the impact of various drivers on sales. Increased sales by 7% as predicted for first quarter
.jpeg?length=1920&name=HB%20Item%202%20(19).jpeg)
