RFID After Marks & Spencer discovered its suppliers, warehouses and storerooms were holding inaccurate records of stock, it conducted a pilot study on item-level RFID tagging
High street retailer Marks & Spencer had been looking atradio frequency identification technology for more than 10 years, but until two years ago, the concept of item-level RFID was thought to be unrealistic.
The technology is still unstable but has improved to the extent that M&S is able to conduct trials based on item-level tagging.
A three-person team worked on a four-week pilot project which ran at the company's High Wycombe store in October 2003. The technology test proved successful, according to Marks &Spencer, with levels of accuracy of between 98% and 100% on fixed and mobile readers.
The company is now planning to extend the trial across more stores and for a longer period so that it can establish a "robust business case".
The retailer hoped the October pilot could reduce the number of mundane activities its staff had to do, such as inventory cycle counting. This would free staff to serve customers and increase sales opportunities.
The pilot was aimed at establishing a level of stock accuracy to reduce the amount of "safety stocks" the warehouses held. The stores were having problems with inaccurate records, deliveries and damage to stock - which was having a knock-on effect down the supply chain.
As a result, before the pilot the stock forecasts were driven by deduced numbers rather than real figures. The only way around this was to carry out manual counts. This took time, was expensive and could disrupt business by removing employees from customer-facing service.
Ideally, Marks &Spencer wanted to keep the minimum of stock, maintained on a real-time basis. Each store can hold six weeks' stock, but extra has traditionally been held in the warehouses because staff mistrusted the forecast information. In addition to this, inadequate forecasting meant that the supplier also held safety stock.
Marks &Spencer did not see the pilot project as a business process change, but rather as a way to allow the company to leap ahead of process advances. If the company could reduce stock from just one location, the project would pay for itself many times over.
The RFID labels were supplied by Paxar, which also provided the labels for Marks &Spencer's clothes. The tags contained a unique identifier which was associated with a catalogue item in the range. No other information was recorded on the tag and no person-based information was held.
For the test, the initial association of tag and garment was made manually by passing the labels across a dual reader once they were attached to the items. In production, the association was made during the garment manufacturing process or in the warehouse.
The paper tags were attached by a plastic connector, which could be removed after purchase.
All the information from the tags was held behind the Marks &Spencer firewall. The tags were 64 bit, UHF 868 MHz chips made by a subsidiary of Swatch, EM Micro Electronics.
The scanner technology was supplied by Intellident, which was already involved with Marks &Spencer through the RFID pilot on its food range, where 2.5 million out of three million items were tagged. However, item-level tagging of food is not likely to be viable for at least three years because of the cost of tags. For the High Wycombe project, less than 25,000 tags were used.
The company does not anticipate that live store roll-out will take place until 2005, even if the project were to be given the go-ahead. The tags used cost 30p, and Marks &Spencer expects it to take at least two to three years for the price to fall to 5p. Item-level food tagging requires the cost to be less than 1p.
RFID business case
Any company embarking on such a project must start with a tightly scoped business case and a vision of the potential - only then does the technology become an issue.
The Marks &Spencer pilot did not need a vast project team. It had one full-time person and three part-time consultants.
Increasing stock accuracy not only reduces costs which, given the complexity of Marks & Spencer's product ranges, can be more than £10m per year, it can also significantly increase sales.
Marks &Spencer has not revealed how much the pilot cost, but the company has shown that item-level tagging will become viable over time. The question it now faces is whether to wait until it is stable and commonplace, or to edge ahead of the competition.
Lessons learned from the M&S pilot
M&S could not buy the handheld devices for the pilot off-the-shelf and instead had to develop them from scratch. The store found that existing devices either had an inconsistent range or a limited battery life. To overcome this, M&S developed the world's first mobile garment reader.
M&S decided not to use intelligent shelving, where shelves are tagged to allow the retailer to track when they need to be replenished. A store's layout changes dramatically depending on the season, and installing an electricity supply for each shelf would be too difficult.
However, the shelf could be tagged and the mobile reader used to read and verify the shelf and its relationship to goods.
M&S found that UHF was the most accurate range to use, even though the store had to set all of its devices to minimum power to ensure regulatory compliance. M&S also discovered that metallic ink used on the tags interfered with the signal.
Nigel Montgomery is director of European research at AMR