So, when faced by an empty shelf, almost half of customers walk out of the store. In fact, shoppers have become less and less patient over the years. While in 1991 only 14% of consumers would have shopped elsewhere as a direct consequence of out-stocks, this figure grew to 31% in 2002, reaching 47% in 2005. Additionally, when disappointed for three times in row, a whopping 70% of customers will walk away – some never to come back.

Despite huge investments in supply chain initiatives and forecasting systems, the problem only seems to get worse. During the ECR (Efficient Consumer Response) conference held in Milan last year it was reported that the average OOS ratio is around 7-10%, with peaks of 30% for products under promotion – this last figure was 13% in 2002.

But the truth is that supply chain initiatives HAVE worked. ECR surveys show high service levels from the manufacturer's warehouse to the retailer's warehouse, and similarly high service levels from there to the retailer's stockroom (98–99%). But this performance drops sharply over the final metres from the stockroom to the shelf (90–93%).

The fact is that ‘supply’ automation must end at the stockroom’s exit.  From that point onwards, on-shelf/on-rack availability is driven by consumer demand – and today demand management is not automated. Consumption levels are usually only known retrospectively by analysing past sales data; this information is then used to calculate the probability of past OOS occurrences and to forecast future consumption. But the reality is that sales data mask out-of-stocks. Typically, 30% of purchases are substitutes for what the shopper originally had in mind; this means that, even if their actual sales data is 100% accurate, retailers and manufacturers never have a truly accurate picture of the true pattern of demand.

Thus, using past sales data for reordering and forecasting purposes inevitably perpetuates out of stocks. In fact, mismatches between forecasts and reality account for 47% of out-of-stocks.

The key to preventing out-of-stocks, keeping shelves full of products shoppers actually want to buy, is monitoring on-shelf availability. But today actual on-shelf availability is monitored visually, i.e. manually, with all the limits of manual activities, e.g. difficult to measure, hard to reproduce consistently, etc. It is not by chance that 25% of out-of-stocks are due to the fact that the product is in the backroom but not on the shelf.

Add to this the lack of visibility for manufacturers over real time availability and sales - according to AMR, this means that it takes a manufacturer a minimum of 2 days to correct an out-of-stock, with an industry average of 7-9 days. This accounts for the final 28% of OOS.

On-shelf and on-rack availability is driven by true consumer demand. The answer to the out-of stock issue then lies in monitoring demand automatically and in real time, across all product categories, and in increasing the automation of shelf and rack replenishment operations.

Such increase in automation is obtained by using each single product ‘reading’- at the point of sale, via a personal self scanner etc. - to instantly update on-shelf availability and sales trends, issuing alerts to operators in case sales trends may lead to out of stocks, and then recording the shelf/rack replenishment operation - either via a mobile device or, for complete automation, with RFID readers at the stockroom’s exit.

The replenishment operation instantly updates both on-shelf and storeroom availability; alerts will then be issued when the storeroom level goes below a pre-defined watermark – and the time between an alert and a shelf or store replenishment is recorded, allowing the measurement of operational performance.

This demand driven approach can be cascaded further down, providing real time visibility to suppliers over sales and stocks by product, store, region and globally.

It is important to remember that a sample based approach, dedicated to monitoring high-margin products, is not likely to work. Customers might be disappointed not find the latest DVD, but they’ll leave the store immediately if they can’t find bread and milk.

The benefits of this demand-driven approach are significant.

First of all, an immediate 1% revenue increase for retailers and $240million more for manufacturers – by eliminating those 25% of OOS due to the product not being visible to customers. Also, instant increase in customer satisfaction and in up-sell/cross-sell opportunities – just imagine shop assistants being able to tell you instantly ‘well, I don’t have exactly what you are asking for, but you should look at this – slightly more expensive, true, but it features in Vogue’s/PCPro last issue – and it goes beautifully with these purple gloves/35’ plasma screen.

Then, real time visibility over consumption trends – i.e. what customers really want and how much they want of it – instantly available to all players within the supply network, with the desired granularity level. Retailers and suppliers will know instantly if customers only started to buy normal milk when the organic range ended, or if that new celebrity dress is selling faster in Oxford St than in Covent Garden, and they’ll be able to adjust supply accordingly – even automatically, by instantly updating delivery schedules and standing orders.