Warehouse Slotting

Travel matters because it is the biggest share of a picker's day: about 50% in the review by de Koster and colleagues, and 55% in Bartholdi and Hackman's textbook. Order picking itself is estimated at up to about 55% of warehouse operating expense in both sources.

A good plan puts busy items near dispatch and at waist height (the "golden zone", where picking is easiest). It keeps heavy items low and avoids packing every busy item into one aisle, which creates congestion.

Slot by pick lines. A pick line is one SKU on one order, and it is what makes a picker walk. The cube-per-order index (COI) formalises this: a SKU's required space divided by its picks per period, with the lowest COI placed closest to dispatch (de Koster et al., 2007).

Sales value answers a different question, which items need tight stock control. The ABC analysis guide covers that ranking.

Here is a real example. The public UCI Online Retail dataset holds a UK gift wholesaler's orders from 1 December 2010 to 9 December 2011. After removing cancellations, non-positive quantities and non-product codes, it held 20,538 orders, 3,919 SKUs and 518,385 distinct pick lines. Distinct orders were counted per SKU and ranked.

The rule of thumb that 20% of products make 80% of activity did not hold here, so measure yours. It is one seasonal wholesaler with long orders (median 15 lines), not an Indian benchmark.

Ranking the same SKUs three ways, with the same filters, gives different top fifths. This table shows the share of all pick lines covered by the top 20% of SKUs (784 SKUs) under each ranking.

This comparison cannot prove that lines are the better measure, because a ranking by lines will always cover the most lines. Lines are recommended because the walk is set by how many trips visit a location, not by how many units leave it (inference). The travel model below tests the line ranking only.

Run six steps in a spreadsheet; each has a pass or fail.

Classes A, B and C by distance from dispatch

1. Collect order lines. One row per SKU per order, covering a full year if demand is seasonal. Pass: every row has a SKU and an order number.
2. Count and rank. Count distinct orders per SKU and sort high to low. Pass: the ranked counts sum to the total pick lines.
3. Cut into classes. Start with three: A for the first 20% of SKUs, B for the next 30%, C for the rest. Research on manual picking finds a full ranking gives the shortest travel but suggests two to four classes in practice. Three classes are recommended because they are simpler to maintain (inference). Pass: class A holds more than half of all pick lines. If it holds less, demand is flat and slotting will gain little (inference).
4. Order by COI inside each class. Divide each SKU's required space by its picks and place the lowest COI nearest dispatch. Pass: no SKU is placed ahead of a lower-COI SKU in the same class.
5. Apply overrides. Heavy items low, busiest items at waist height, hazardous goods segregated, expiring lots reachable first, and no aisle holding every A item. Pass: no heavy SKU sits above waist height and no aisle holds more than half of the A SKUs.
6. Time it. Time at least 50 orders before and after the move (suggested), using the method in the order picking methods guide. A mispick is the wrong item or quantity picked. Pass: the average time per order falls and the mispick rate does not rise.

Re-rank at least every quarter, and sooner after a launch or season change (recommended). In this dataset, of the 675 SKUs in the first six months' top fifth, 438 (64.9%) were in the top fifth in the following six months.

The model used one aisle, one slot per SKU numbered by distance from dispatch, and one trip per order to its farthest slot and back. Random placement scores 1.00 (average of ten random layouts).

The same-period column slots and tests on the whole year. The later-period column slots on orders to 31 May 2011 and tests on 1 June to 9 December 2011. SKUs unseen in the first period were placed last. Part of the gap between the columns is overfitting (the plan is fitted to the orders it is tested on) and the autumn peak, which falls in the later window.

Also, 16.2% of later pick lines were for SKUs absent from the first period. The model is a distance proxy, not a route simulation, and it tests ranking by lines only, not COI. Move only SKUs that changed class. Bin records must also be right, so pair slotting with cycle counting, which means counting a small part of stock every day or week to correct bin records.

Slotting is a plan; putaway is the act of storing received stock, which should follow the plan. Retail "slotting fees" are one-time payments a supplier makes to a retailer for initial shelf placement or warehouse access, a different meaning. Some texts use slotting for case orientation and shelf placement.

Related terms: pick face (the location items are picked from), golden zone, re-slotting (moving SKUs after re-ranking), replenishment (refilling pick locations from bulk storage), COI and ABC analysis.

A slotting plan is only as good as the bin and pick records behind it. In our warehouse management feature you receive stock by barcode, put pallets away into bins and generate pick tickets that suggest the bin to pick from. Lot-enabled products are picked expiring-first, and other items can follow first-in, first-out. Pallet labels and multiple warehouses are supported, and custom reports and dashboards let you build your own views. We supply the bin and pick records; the ranking itself happens in your spreadsheet, using the sheet above.

Our 15-day free trial gives full access with no credit card.