Replenishment algorithm logic

Nextail replenishment algorithm calculates the stock quantities to send to each store maximising sales probability across the network. It is based on guiding principles that help to understand how it works and why.

Demand (not sales) forecast

All calculations are based on the sales probability of an item in each store each day.  Why?

• Past mistakes in inventory allocation are not “carried over”.
• The system corrects and actively prevents stockouts.
• The full potential of the sales network is optimized.

Nextail forecasting module

Global (not local) optimization

The replenishment cycle is calculated as a whole, trying to achieve an optimal result. Why?

• No local optimisation, no concern for “store level goals”.
• Each unit is sent where it is most valuable (considering cost and constraints such as minimum display, packaging size for distribution, the distance between each store and warehouse, etc).

Optimization at product/sku level

Robustness over accuracy

Avoiding large mistakes takes precedence over increasing accuracy in most cases. Why?

• The system works well even with inaccurate or scarce data.
• Making the system robust ensures the best results in the long term, while it may sacrifice small potential short-term gains.

Meritocracy

The replenishment calculation start point is the existing stock, not store demand. Why?

• No "competition between stores for the same item”.
• Each stock unit’s performance drives its own replenishment.

Rich constraints set

Several types of constraints can be configured and considered. Why?

• “Constraints” allow the input of many business requirements when making the calculations.