Launching a new product without out-of-stocks: how to order a SKU with no sales history?

A product the system does not know

A SKU’s reorder point is calculated on its daily sales and its delivery lead time (automatic ordering and out-of-stocks). For a new product, daily sales do not exist. The system gets a default value, or whatever someone typed when the record was created, and orders on it for weeks.

Three launch specifics add to that:

• concentrated demand: advertising, leaflet, gondola-end display (leaflet and availability);
• a facing decided before sales, often too short if the product works (facing and shelf capacity);
• a first delivery sometimes incomplete or late, the supplier ramping up too (receiving and checking by SKU).

Why a launch out-of-stock costs more

Facing an out-of-stock on a product they know, the shopper substitutes, delays or goes elsewhere: Gruen, Corsten and Bharadwaj measure 15% delaying. For a product they have never bought, there is neither habit nor attachment: if they do not find it the first time, they have no reason to look for it a second. The store does not lose a sale, it loses the chance to build a buying habit. And it draws the wrong conclusion: “the product does not sell”, when it was not on the shelf.

The method

1. Set parameters by analogy

Choose an existing SKU that resembles the newcomer (same family, same price bracket, same brand if possible) and take its sales as a base, corrected for advertising support. It is an estimate, but it beats a default value.

2. A launch stock in the back room

For the first two to four weeks, an identified back-room stock that the team puts out by hand, without waiting for automatic ordering to react (organising the back room).

3. A daily count for four weeks

The new SKU joins the list counted every day, like a promotion: present or absent, at both locations if it is displayed twice. That count is the only way to tell “does not sell” from “was not on the shelf”.

4. Check the first deliveries by SKU

The record has just been created: errors in code, packaging or units per case are frequent at the start. An error at the first receiving skews book stock from day one (phantom stock).

5. Recalculate on real sales

After four weeks, sales exist. Reorder point, maximum stock and facing are recalculated, excluding from the average the days the product was absent, otherwise the system learns the out-of-stock and orders less.

Week	What drives the order	What the store does
Before launch	estimate by analogy	record created, launch stock, provisional facing
Weeks 1 to 4	estimate, corrected by hand	daily count, replenishment from the launch stock
Week 5	real sales, excluding out-of-stock days	new reorder point, adjusted facing
Then	automatic ordering	normal tracking

The skewed-average trap

This is the least known point. If the new product was out of stock five days out of twenty during its launch, its average sales are understated by a quarter. The system derives a reorder point that is too low, which creates new out-of-stocks, which lower the average further. The daily count also serves that purpose: knowing which days to exclude from the calculation.

A worked example (hypothetical)

Take a fictional supermarket launching a new drink with a facing of two and a default value of four units a day. Supported by advertising, it sells about a dozen a day when present, and runs out every afternoon. Over four weeks the system records an average of six a day, because the shelf was empty part of the time. With parameters set by analogy at ten a day, a launch stock and a facing of four, the expected scenario is a product present all day, and a real average measured around twelve. The figures are illustrative.

What shopper reporting changes

A shopper who takes the trouble to report a new product missing says two things: the product is absent, and it was expected. With ShelfAlert, the store knows this in the first days of the launch, before automatic ordering has learned the wrong figures. 14-day trial, no payment card: see the plan for store managers.

Sources

• Gruen, Corsten, Bharadwaj, “Retail Out-of-Stocks: A Worldwide Examination”, GMA, 2002: full study (PDF).
• Corsten, Gruen, “On Shelf Availability”, 2004: summary.