Receiving and checking by SKU: how to prevent out-of-stocks from the loading dock?

What checking by case lets through

Most stores receive deliveries by counting cases and signing the delivery note. The system then records ordered quantities as delivered. Anything that differs inside the cases becomes a book stock error:

• a SKU missing from a case that looks complete;
• a SKU substituted by the supplier or the warehouse (other flavour, other size);
• transport damage;
• a mix-up between two stores on the same round.

DeHoratius and Raman measured 65% of inventory records inaccurate at a retailer, and part of those discrepancies is born at the dock. The mechanism is described in phantom stock: what to do when the book stock says the product is there: stock shows the product, automatic ordering does not fire, the shelf stays empty until someone notices.

Check by SKU, not by case

Checking by SKU verifies that each expected SKU is present in the expected quantity. On a delivery of 60 cases and 400 SKUs, doing it in full takes an hour: rarely possible. The method is to target:

1. The store’s fifty best sellers, at every delivery: they cost the most when out of stock.
2. Promoted products in the week of the operation: an incomplete promo pallet is a promotion out of stock from day one (promotions and out-of-stocks).
3. Small expensive products (pharmacy, razors, cartridges): frequent errors and disappearances.
4. A rotating sample of ten SKUs per delivery, to detect problem suppliers or rounds.

On these SKUs the check is unitary: the case is opened, the content counted, the discrepancy keyed into the system before book stock is updated.

Record the discrepancy, not just notice it

The check is useless if the discrepancy does not enter the system. Three rules:

• the discrepancy is keyed the same day: book stock must reflect the real delivery before the next day’s order;
• the discrepancy is dated and attributed (supplier, warehouse, round): after a month, the discrepancy table names the cause to treat;
• the back room receives the product in its slot (organising the back room to avoid out-of-stocks): a product checked then lost in the back room recreates the out-of-stock just avoided.

Reading the discrepancy table

After a month, the receiving discrepancy table answers three questions:

Finding	Likely cause	Action
Discrepancies concentrated on one supplier	supplier picking	dispute, systematic check of that supplier
Discrepancies concentrated on one round	mix-up between stores	exchange with the warehouse and the neighbouring store
Discrepancies on small expensive products	disappearance between warehouse and shelf	sealed cases, check on opening
Positive discrepancies (more delivered than ordered)	order keying error	check ordering parameters

This table also feeds the store’s real supplier service level, often different from what the warehouse announces (supplier service level and out-of-stock rate).

A worked example (hypothetical)

Take a fictional case. 2,500 m² supermarket, one warehouse delivery a day. Receiving by case, no discrepancy recorded. Cycle counting set up on 50 SKUs finds 21 overstated book stocks in a month, 9 of them caused by an undetected incomplete delivery. After moving to checking by SKU on those 50 SKUs and on promotions (25 minutes per delivery), the following month counts 4 receiving-related discrepancies in cycle counting, and reported out-of-stocks on those SKUs fall by a quarter. The figures are illustrative.

What shopper reporting changes

A product reported absent by shoppers on a delivery day, while book stock has just been credited, is the first sign of a wrong delivery. ShelfAlert gives that list the same evening: the next morning’s check knows where to look. 14-day trial, no payment card: see the plan for store managers.

Sources

• DeHoratius, Raman, “Inventory Record Inaccuracy: An Empirical Analysis”, Management Science, 2008: abstract.
• Corsten, Gruen, “On Shelf Availability”, 2004: summary.
• Gruen, Corsten, Bharadwaj, “Retail Out-of-Stocks: A Worldwide Examination”, GMA, 2002: full study (PDF).