Article - Issue 18, February/March 2004
High Street: The new technological battleground?
Modern technology employed by the retail industry already profiles customers for targeted advertising and records our every purchase. The latest advances include video screen modelling in changing rooms and the prospect of having every item in a basket scanned simultaneously at the checkout. Allan Dean explains loyalty cards, basket analysis and radio frequency identification in retailing.
Compared with the technical complexities of telecommunications or construction, the retail industry seems to have little to tax the brains of engineers. But retailing is a competitive business – arguably the most competitive. New technologies and business practices are starting to provide major retailers with a chance to gain an edge over their rivals.
The rise of hyper-competition in retailing
Retail is a deceptively simple industry. It is a matter of buying items in bulk and selling them to the general public, ideally for a profit. Historically, the key to profitable retailing was to build customer loyalty though good service, while selecting the most profitable products to display in each square foot of floor space. The major challenge was to judge demand correctly for any particular product; all stock has to be sold somehow, and no retailer wants to be forced to sell at a loss to release floor-space for more profitable lines.
As a result, talented buyers (who had an intuition for the public demand for a product) were key, and generally found the fastest route up the career ladder at the expense of ‘hard skilled’ engineers.
This resulted in a culture where manufacturers and importers competed for the attention of the retail buyers, and supplied the bulk of the innovation in the industry. Retailers generally found technologies and processes that worked and stuck with them.
Large suppliers were able to achieve economies of scale that displaced smaller players, and this resulted in the rise of the FMCG (fast-moving consumer goods) manufacturers like Unilever and Rank Hovis McDougal in the 1960s, the flood of electronics goods from the Far East in the 1980s and the popularity of designer clothing brands in the 1990s.
Gradually, retailers found their competitors buying from the same suppliers and selling to the same customers, leaving the selling price the only difference in their offering. Competing on price alone is a very uncomfortable prospect, since in the long term it reduces everybody’s profits to the bare minimum. Some other means of differentiation was required.
Tesco – the return of the individual
The major supermarkets were one of the first sectors to find themselves forced to compete on price, with a number of ruinous price wars through the 1970s and 1980s. Supermarkets struggled to find ways to squeeze an extra one per cent or two of margin, which is very difficult when the major players all provided a very similar product. Tesco scored significantly against the then leader Sainsbury in February 1995 with the launch of its Clubcard scheme.
The rationale for loyalty schemes dates back to Green Shield stamps in the 1950s. By encouraging customers to save up points for free gifts, customers were dissuaded from shopping with rival retailers who were not part of the scheme. But the great innovation of the Clubcard was that it was operated through the point-of-sale system. Back at head office, Tesco’s buyers, marketing managers and store planners could see exactly who was buying what. A bevy of customer relationship management (CRM) technologies could then be used to further increase margins:
Basket analysis looks for patterns in items bought in the same visit. Statistical analysis finds ‘affinities’ between products, some of which can be surprising at first. It is not clear why beer and nappies tend to appear in the same shopping basket, but it could be a result of working fathers picking up emergency supplies of both on the way home. By placing items with affinities close together on the shelves, total sales can be increased. A similar technique spots stores that sell an exceptional quantity of a certain product, and these stores are examined to find techniques that can be replicated elsewhere.
Customer segmentation categorises customers according to the pattern of items they buy. Tesco has identified more than 3000 such patterns and uses a customer’s buying history to identify, for example, that a customer fits the profile of a single man with a good salary, a cat and an enthusiasm for curry. By examining the purchases of other customers in the same segment, advertising and offers can then be targeted towards customers who are likely to respond positively.
Life-stage analysis looks at patterns over time. It picks out purchases that indicate forthcoming events like the birth of a child or a new job. These life changes result in customers buying classes of products for the first time and are an excellent time to build loyalty. Tesco has the largest and most profitable grocery website in the UK. This is due partly to the integration between the website and the Clubcard system. The special offers, advertisements and recommended items the customer is shown on the website (‘upsells’ in retail jargon) depend on the customer segments to which they belong.
A significant benefit of the Tesco implementation of the loyalty card is the simplicity of the technology, resulting in much lower overheads than similar schemes introduced by other UK retailers. Each Clubcard contains a random, unique identifier pre-encoded on an ISO 7811 magnetic stripe. This is the same standard used for credit cards, so no additional hardware is needed at the point of sale. Just one additional data field – the Clubcard ID of each purchase – needs to be added to the existing overnight submissions from the store to head office.
Customers can pick up a card instore at any time. New cards are glued to a form (a ‘header’), which is pre-printed with the card number and contains spaces for the customer to fill in their name, address and other useful demographic data. The customer can start using the card (and supplying Tesco with data) immediately, but it may be several weeks before Tesco receives the header and finds out the real identity of the shopper.
Recording every purchase by every customer generates a lot of data, and Tesco again took a very conventional approach to the problem by extending the IBM MVS mainframe environment it was already using for managing store logistics. While this was an unfashionable approach at the time, it minimised the complexity and risk in the project and allowed Tesco to run industrial-strength data analysis tools (principally supplied by SAS) on the data.
The downside of this approach is that all data is held centrally and processed in overnight batches. The only information generally available in-store is the card number. To offer targeted promotions to selected customer segments, Tesco sends the customer money-off vouchers through the post. These vouchers include a bar code which triggers discounts on selected products when scanned at the till. The accuracy of Tesco’s customer segmentation ensures that 80% of these vouchers are redeemed. This is a very impressive figure – response rates of 2% are more typical. Since many of the offers are jointly funded by the manufacturers the profitability of the scheme is obvious.
While it may be considered inelegant for customers and staff to handle paper coupons, the success of the scheme cannot be questioned. Unlike conventional coupon-based promotions, the coupons can be destroyed as soon as they are scanned, avoiding the significant cost of handling coupons in the store’s cash office. Alternative technologies have been used by other retailers. The Boots Advantage card stores the customer information on a chip-enabled Smart Card, but installing a Smart Card reader at every till significantly added to the cost of the £25 million scheme. The WH Smith Clubcard, based on the Fujitsu Corema platform, offers the greatest flexibility: the customer’s complete profile, including any targeted offers, is retrieved in real-time from the central database every time the card is used. While both these schemes have yielded valuable data for the buyers at head office, nobody can match Tesco’s skill in targeting individual customers.
The importance of the supply chain
The fashion industry went through a tough time in the 1990s. Mid-market retailers found that they lost market share to cost-competitive newcomers like Asda and Matalan, while aggressively marketed designer brands poached the more discerning customers.
Retailers, even the famously patriotic Marks and Spencer, consolidated production with large Far Eastern manufacturers. Besides taking advantage of cheaper labour costs, concentrating orders with fewer suppliers is a common approach to reducing non-quality costs and inventory in any manufacturing process. However, the complexities of dealing with the Far East caused the lead time for designing, manufacturing and delivering a new range to rise to about nine months. Unfortunately, the UK population doesn’t always follow the fashion industry’s plans, and the UK climate is similarly unpredictable. Once the supply chain is running, the buyers have little to do but bite their fingernails.
An alternative approach has been taken by the Spanish fashion retailer Zara. Rather than extrapolating from catwalk shows to predict fashion trends nearly a year in advance, Zara can watch for trends that have already taken hold among trendy urbanites. They then design a new range of clothes in response, and have them in clothes stores within two weeks.
To achieve this level of market responsiveness, designers and production planners have direct access to sales data from the point-of-sale system in the stores. They know what is selling well and can manufacture more products in a similar vein. To supplement this, each store manager is issued with a hand-held device that they use to record the reaction of customers to new ranges, and to capture emerging fashion trends in the dress of their clientele. Again, this information is available for analysis within 24 hours. The result is a supply chain which, conceptually at least, starts with the customer.
Once new designs have been agreed, cloth is cut automatically using Zara’s own highly automated pattern-cutting machines working on small batch sizes. Conventional wisdom would dictate that the garments should be sewn and finished on a similar, heavily industrialised scale. However, Zara take a very different approach. They have partnerships with a wide range of small-scale workshops in Northern Spain and effectively configure a new supply chain for each batch. These workshops are too small to have their own computerised production management systems, so Zara supplies them with this technology. In return, Zara gains visibility of their supplier’s activities. In effect, they have converted their supply chain into a very flexible ‘value net’, through which products can take a wide variety of paths. Sending small batches through a short supply chain has obvious advantages in reducing stock, which translate directly into efficient use of working capital.
The success of this approach is obvious. In many European cities, ‘Zaramania’ breaks out among the locals on scheduled delivery days. In an industry whose profits are very sensitive to the amount of unsold stock at the end of a season, the financial benefits are enormous. Zara has more than doubled its size in the last five years, and its profit margin of 27% leads the industry.
Prada’s high-tech customer service
Prada is another retailer using IT to differentiate itself, in its case from the crowded luxury goods market. E very garment in the Epicenter, the New York flagship store, carries a radio frequency identification (RFID) tag (see sidebar). Staff and regular customers use hand-held devices to read these tags, allowing them to check stock levels and assemble customer orders. The same tags trigger LCD screens in the changing rooms to play video clips modelling the same clothes that the customer is trying on, while the mirrors have been replaced with video cameras and plasma screens. These ‘mirrors’ buffer the video stream for five seconds, so if a customer turns around quickly they can see themselves from behind.
In the retail IT industry, these innovations have made the Epicenter the most famous store in the world, and guaranteed a steady stream of engineers and academics coming through the doors. However, it is not clear to what extent they have increased sales in their target market, and anecdotal evidence suggests that usage of the technology is low.
Achieving advantage through IT
Tesco and Zara are leaders in their respective sectors, and even their direct competitors will privately acknowledge that they are the retailers to beat. Prada, whatever the direct effect of its technical innovation on its sales, has received publicity for the Epicenter that eclipses its rivals.
While these successes cannot be ascribed entirely to technology, these retailers share a common approach. They have all adopted technological innovation as an integral part of their competitive strategy. By treating IT as an enabler, they have taken a direction that sets them apart from their rivals. If this trend continues, and other retailers retaliate in kind, retailing may hold a lot more challenges for engineers in the years to come.
Radio frequency identification
RFID (radio frequency identification) is a relatively new name for a mature (if revamped) technology. Historically, RFID has been used to identify items over short range. One example is in the implanting of chips in pets and livestock, allowing quick identification by radio using a hand-held reader. A number of new developments have brought RFID to the attention of IT directors in the retail industry.
RFID transponders (tags) can be active or passive. In active RFID, a battery-powered transponder is activated by a reader, which emits low-power radio field at a specific frequency, typically 125 Khz. This triggers the transponder to emit a series of radio pulses in a time-coded pattern unique to the transponder. It is possible to identify a transponder at a range of tens or hundreds of metres, but the cost of the transponder and the need to replace batteries limit the useful applications of active RFID.
Passive RFID does not contain a power supply in the tag, which consists simply of a small silicon chip and a copper-coil antenna. The radio signal emitted by the reader induces a current in the tag’s coil sufficient to power the device. When powered, the transponder repeatedly shorts out its own antenna causing a detectable change in impedance in the reader’s antenna. Each tag uses a unique sequence of pulses, allowing the tag (and hence the item to which it is attached) to be identified. Passive RFID tags can be read at a range of 10 to 30 cm using a hand-held reader, or by passing them through a loop-shaped ‘gate’ antenna up to 2 m in diameter.
Even though a frequency of 125 KHz became the de facto standard for RFID, manufacturers historically tended to implement proprietary standards for RFID, since they often used ongoing revenue from tags to subsidise the initial cost of the readers. The new ISO standards for RFID have opened up a large number of new opportunities for RFID technology, which has in turn encouraged manufacturers to adopt the standard and provide interoperability between tags and readers from rival firms. ISO RFID cards use the higher frequency of 13.56 MHz and offer a number of advantages:
the antenna to be reduced from a hundred or so turns of wire to about ten, which can be cheaply manufactured as a printed circuit on a flexible plastic substrate.
higher frequencies and power allow more sophisticated microprocessors to be used, allowing secure, encrypted transmissions and protection from duplication.
to avoid ‘collisions’, each tag and reader waits for a random length of time between transmissions, so a tag need not be removed from the field before another can be read, and readers do not need to be kept well separated.
the sensor can modulate its own field to transmit blocks of a few kilobits of information for storage in the tag; multiple blocks allow the tag to be used for more than one purpose.
ISO RFID tags currently cost from as little as 30p, and this price is predicted to fall below 1p over the next five years.
This raises a number of possible applications. ISO RFID cards have already replaced magnetic stripes in season tickets on the London Underground. In the retail industry, Visa and Mastercard have announced their intention to embed ISO transponders in credit cards, and at least one retailer is examining the possibility of identifying customers while their loyalty card is still in their pocket.
However, the biggest impact of RFID is in the retail supply chain. With RFID labels potentially costing little more than conventional swing-tags, every item of stock can be uniquely identified at a distance. An item’s life history can be written incrementally to the tag at each stage of the supply chain and carried from company to company with the item. This is much simpler and cheaper than integrating incompatible supply chain software with multiple suppliers.
Most significantly, a carton containing hundreds of items can be passed through a 2 m diameter coil antenna, counting and uniquely identifying every item in a couple of seconds. The greatest element of administrative burden in the retail supply chain is double-checking physical stock against electronic records and dealing with discrepancies. The potential savings to be derived from automatically and accurately counting items at every stage of the supply chain are enormous.
Next Door Consulting
Allan Dean is a co-founder of Next Door, a consulting firm that collectively acts as an ‘on demand’ Technical Director to mid-sized companies. Originally from a management consulting background, he became eCommerce Manager for the online arm of Arcadia Group, the UK’s second largest fashion retailer during the dotcom boom. Here he was responsible for the well-being of 22 websites including Top Shop, Dorothy Perkins and Evans, and the associated logistics for transporting garments to customers and back. He was subsequently responsible for Retail in the e-Business division of ICL (latterly Fujitsu), the largest supplier of IT services to the UK Retail market.