How to measure sales lost to out-of-stocks?: ShelfAlert

In shortLost sales are calculated by comparing, for each reference out of stock, what it would have sold during the out-of-stock (its usual sales velocity at that hour) with what it did sell, then keeping only the share really lost for the store, about 55% according to measured shopper reactions. You therefore need three things: which references were missing, for how long, and how fast they sell.

Why the figure exists nowhere

The accounts record what was sold. An out-of-stock is a sale that did not happen: it leaves no trace, except an unhappy shopper nobody saw. That is why shrink, visible at stocktake, is handled rigorously, while out-of-stocks, invisible, are underestimated. The ECR France / IRI barometer nevertheless puts the lost sales of out-of-stocks at more than a billion euros a year for French grocery retail.

The three pieces of data you need

To estimate a lost sale, you need:

The reference out of stock: which product, in which aisle.
The duration of the out-of-stock: between the moment the facing emptied and its refill.
The usual sales velocity of that reference at that time slot: units sold per hour on a comparable day.

The formula for one out-of-stock: gross lost sales = sales velocity × duration. Then apply the share really lost for the store. According to the worldwide study by Gruen, Corsten and Bharadwaj, 31% of shoppers buy elsewhere, 9% give up and 15% delay: counting delays for half, about 55% of the sale is lost, the rest being recovered through substitution.

Method 1: estimate from the out-of-stock rate

The simplest, for a store-level order of magnitude: turnover × on-shelf out-of-stock rate × 55%. A supermarket at €60,000 a day and 7% out-of-stock loses about €2,300 a day. It is rough (references are not all equal) but enough to decide to act. The detail of the calculation is in how much does an out-of-stock cost a store.

Method 2: abnormally low sales

More accurate, it starts from till data: for each reference, compare the day’s sales with the average of the same weekdays of previous weeks. A reference that sells 40 units on an ordinary Tuesday and 12 this Tuesday, with no promotion or price change, very probably was missing part of the day. The gross lost sales are the difference (28 units). This method detects out-of-stocks after the fact and does not say at what hour they happened.

Method 3: direct measurement

The most accurate: knowing the start and end time of each out-of-stock. It assumes continuous detection, by shelf sensors, cameras or shopper reporting. You then multiply the real duration by the hourly sales velocity of the reference. It is the only method that lets you say “semi-skimmed milk was missing from 5:10 pm to 7:30 pm on Tuesday and Thursday, and it cost 90 units”.

Example on an aisle

Dairy aisle, reference semi-skimmed milk 1 L, sales velocity 36 units an hour in the evening. Out-of-stock observed from 5:10 pm to 7:30 pm, that is 2 h 20.

Gross lost sales: 36 × 2.33 ≈ 84 units.
Share really lost: 84 × 55% ≈ 46 units, about €50 of turnover on a single reference, a single evening.
If the out-of-stock repeats three evenings a week: nearly €8,000 a year on this one reference.

Multiplied by the twenty fast movers of an aisle, you understand why evening out-of-stocks weigh more than the store average lets you see.

What shopper reporting changes

ShelfAlert provides the data method 3 lacks: every report gives the reference, the time and the store, and the dashboard shows the most-reported products and the slots when they are missing. The manager can then put a figure on lost sales from their own data, aisle by aisle. 14-day trial, no payment card: see the plan for store managers.

Sources

Gruen, Corsten, Bharadwaj, “Retail Out-of-Stocks: A Worldwide Examination of Extent, Causes and Consumer Responses”, GMA, 2002: full study (PDF).
ECR France / IRI out-of-stock barometer: LSA, out-of-stock rate and Les Marchés, on-shelf out-of-stocks.

How to measure sales lost to out-of-stocks?