Retailers and manufacturers need forecasting in all facets of their organization to support decisions being made by merchandisers and category managers, ad planners and marketers, but also by buyers, replenishment and vendors in collaboration situations - for their planning and operational processes. AMP2’s rich data sources ensure the elasticity demand models encapsulate real insight into the drivers behind consumer behaviour and demand. AMP2 creates a non-linear forecast using dynamic models and advanced algorithms to understand and predict future sales in fast, medium, slow and very slow moving retail environments taking account of the step changes in demand that occur due to price changes and promotions etc.
The AMP2 tool set provides data cleaning, model building, model update and forecast accuracy facilities as well as new item/SKU introduction and item deletion without models having to be rebuilt from scratch. The AMP2 models are constantly updated and refreshed and do not degrade as new sales and customer demand is updated each day or week. The majority of forecasting solutions used by retailers are based on linear algorithms which break down when price, promotions, ads etc. occur and cannot track the change in the demand uplift or revert-back phase of the promotion period: AMP2 can and does. AMP2 also tracks the impact on related products which are affected by a price change or promotion, cannibalization-halo effects etc., Accurate forecasts will provide benefits all round in better promotion planning, fewer out-of-stocks, improved service levels and reduced safety stocks throughout the whole supply chain infrastructure.
AMP2 forecasts can be relied upon and used with confidence to make important business decisions
At the heart of the solution is a demand model calibrating engine capable of calculating price and demand elasticity from past sales data. Using weekly or daily sales information it uses extensive data cleaning techniques to calculate missing or unstable values. The solution automatically determines the best mathematical forecasting algorithm to use for each item based on goodness-of-fit and other advanced statistical performance metrics.
The major components of the Forecasting module are: