Case Study · Retail
Optail Replenishment: automatic warehouse-to-store feed for what sells
Replenishment, live since October 2024, manages the warehouse-to-store flow by rate of sale: how many weeks of cover, minimum lot, which warehouse, which store group. In the first two months two thirds of replenishment came from Optail; in 2025 it approached 95% in footwear. Whether every replenishment turns into sales is measured — Optail's sell-through rate is higher than manual and than last year.
- ~95%
- Automation
- ~1.5×
- Sell-through
- 25%
- Fewer shipments
- 2
- Parameters
Footwear 2025; ~70% in the first two months
Optail vs manual, winter season footwear
For roughly the same sales, first season
Weeks of cover · minimum lot
Problem
If a product that sells in a store is not replaced, the shelf stays empty; if stock is sent to a slow seller, the warehouse drains, the fast-selling store waits while goods pile up in the slow one. At Derimod this decision was made every day, store by store, by the planners — and once made, whether that shipment turned into sales was never tracked.
How it works
Replenishment is Optail’s second module, built on the transfer infrastructure. Same scenario logic: the user picks the algorithm and its type, enters gender, main category and season group, sets the sending warehouse and the receiving store or store group. Transfer had receiver and sender cover; replenishment needs two parameters: how many weeks of forward cover and minimum lot.
The algorithm runs on current sales and stock in the data warehouse: for every store–SKU it computes a target stock from the rate of sale, subtracts on-hand and in-transit stock, and meets the need from warehouse stock. When the warehouse is short, the higher-potential store takes first. The result is at store × product × size level; the approved proposal goes to AX as a dispatch order.
The special cases screen changes or excludes the rule at any level of the hierarchy: the list of products not to be auto-replenished, specific store groups, separate behaviour for web sales. Run orders schedule the regular scenarios; when the planner arrives in the morning the proposal is ready. A run order cannot hold two different algorithms — replenishment and transfer are fed by different procedures and are not mixed.
Measuring
What does it mean for a replenishment to be “successful”? We set the definition up front: the first inbound movement is linked to the first outbound movement. If 1 of the 2 units sent to a store sells within the window, that shipment’s success is 50%. The same measurement is made separately for initial allocation, replenishment and transfer; separately for Optail and manual; and for the same period last year (LFL).
Results as presented to the board, rounded:
| Period | Automation | Sell-through · Optail | Manual | LFL |
|---|---|---|---|---|
| Oct–Nov 2024 · footwear | ~70% | 30% | 20% | 24% |
| Oct–Nov 2024 · bags | ~45% | ~50% | ~28% | ~36% |
| Winter 2025 · footwear, last 3 months of season | ~95% | ~46% | ~30% | ~36% |
| Winter 2025 · bags | ~50% | ~63% | ~36% | ~50% |
In the first season automatic replenishment created about 25% fewer shipments than the manual period and turned into roughly the same number of sales. Comparing the most manually overridden stores with the most automated ones, the automated group led in units sold — while footfall fell.
One example: unlike the previous year, when stock was pushed to stores in bulk before the November sale and then supported only sparingly, in 2024 shipments went day by day to the stores with the highest potential; the sales and stock curves ran in parallel.
Why this way
- The success definition before the algorithm. Without a measurement, “is automatic better” is an argument of opinion; the first-in–first-out link turned it into a number.
- Few parameters. Weeks and minimum lot; the rest in special cases. The planner understands two numbers and owns them.
- The automation rate is an outcome, not a target. The rate rose on its own as planners stopped correcting proposals; the corrections became rules.