Cards have become more important than cash in moving around London. In July 2014, Transport for London (TfL) stopped accepting cash on its buses, telling commuters they had to use Oyster cards, or debit or credit cards with a contactless payment chip. It also began to accept cards on its Underground and rail services.
TfL flagged up the customer convenience angle and acknowledged the change would make big savings over time – but the move to card payments will also provide more data it can use for long-term planning.
It demonstrates the potential to tap data from near field communication (NFC) cards and radio frequency identification (RFID) tags for analytics at macro and micro levels. The current norm is for RFID middleware to filter the data it reads into an enterprise system to monitor activity; but it is also possible to direct it towards a business intelligence system for analysis. There is even a basic analytical capability in some middleware, such as Checkpoint’s OAT Foundation Suite and Xterprise Clarity.
It is still early days but, as people make more small payments by swipe card and companies attach RFID tags to more goods, there will be a torrent of information on customer behaviour. This is going to prompt more organisations to look at how they use that data.
TfL has been gleaning information from Oyster cards for 12 years. Sashi Verma, its director of customer experience, says the cards provide data for “demand profiles”, showing how many people enter and exit the transport network at specific points, and where the peak load develops on any route. They can tell TfL where people get on buses, and where they get on and off trains.
Read more about RFID data analytics
This makes it possible to construct a picture of individual journeys and how the bus and train networks are used by the public.
“The Oyster Card advantage over survey data is that the information is much more granular,” Verma says. “I can conduct any kind of demand analysis for any day of the year.
“If you want an analysis for yesterday I can give it to you today. If you want analysis for unusual events we can study those across the network.”
He says it is possible to create any number of metrics.
“Every time you touch your card we get the card number and time and location of use, from which you can construct any metric – the demand profile of a particular bus stop or station, the load profile of sections of train lines or bus routes, the off-peak profile of demand, the day-by-day profile over a week, seasonal profiles, and so on. There’s no shortage of what you can construct.
“The point is that the underlying data is so granular you can construct an image of almost any analysis you want on the transport network. You don’t need to know a whole lot more. All of those things – frequencies, design of bus routes and interchanges, operational procedures – there’s no end to the kind of work you can do. It can inform every level of development.”
Verma emphasises that this works with anonymised data, but says knowing the identity of a customer could be useful for future services – such as giving them information when a station is closed.
RFID analytics in retail
The question of how identity can be combined with RFID is already being addressed in retail. According to Stephen Gallagher, vice-president of business intelligence at IT services supplier Avanade, the industry has been more ambitious than most in analysing data from payments cards and RFID tags on goods for sale. The company has worked with retailers on using RFID in supply chain efficiencies and fraud detection and, more recently, has begun to try to better understand customer preferences.
“If you have vast sets of data you have a better knowledge of customer preferences, which helps in terms of manufacturing,” Gallagher says. “You get better inventory and material tracking to provide supply chain efficiencies.
“It helps with theft prevention and with shrinkage – losses to the supply chain – and helps us to become more efficient in terms of the supply chain and the customer. RFID can better identify where those losses are occurring in the supply chain, in the warehouse, transit or in the store.”
But it’s the move towards one-to-one marketing, where the company can learn from how an individual has behaved in a store, that is opening new possibilities.
Tracking customer behaviour
For clothing, it’s possible to place sensors in a fitting room to read RFID tags and measure how many are taken off the racks and for how long, and to record the groups of items being tried by a customer. This can be used to provide an overall picture of activity in the store, identifying if customer behaviour varies at different times, or targeted at an individual.
Avanade has used the technology in its Connected Fitting Room, which is used in the Kohl’s clothes store chain in the US. The sensors read the tags, the system matches the items with others and makes recommendations on a screen. The customer can then indicate if they want a store assistant to bring any of the recommended items to the changing room.
“Although they will be anonymous to us at that stage, we can start to treat them as an individual customer,” Gallagher says.
“When they make a purchase we can put an identity on them. This ties in with customer loyalty schemes, and we can get more information than just what they have purchased, with buying and non-buying behaviour.”
He points to a possible use of this where, if that customer has a loyalty card and logs on to the retailer’s website in the future, the home page could be tailored to match the analysis of that customer's taste.
Gallagher acknowledges the privacy concerns, but says RFID tags only provide a record of the customer’s movements around the store after they have picked up an item and before they leave. For any purchase, the tag becomes inactive once it’s taken outside.
RFID analytics in mobile
Paul Crutchley, strategic engagement director at GSMA, the international association of mobile operators, highlights the potential for understanding customer behaviour. He says the most important thing in using RFID with analytics is to understand the customer journey.
“It’s not so much about the intricacies of the data and how you want to supply it; you have to work out how you supply it to match the customer journey,” he says. “For example, in tracking a parcel you have to ask: ‘What is its journey, who needs to interact with it, and where and for how long do they interact?’”
He says real-time analysis aimed at the individual customer is most effective when it incorporates contextual data, which can be supplied through mobile devices. For example, you can inform the customer of where they can buy an umbrella if the weather is bad; or give them a different travel option if their train is cancelled.
RFID will be most valuable in analytics when data from different sources – such as application developers, Wi-Fi providers and brand suppliers – is combined.
But there is a problem with both these possibilities: The different players involved can have different demands, and there is no agreement on the key interaction points and how they should be measured.
The GSMA has been working with other organisations to establish some standards to support interoperability. But Crutchley says progress has been slow, hampered by the fact that “innovation often creates fragmentation”. Compounding this is the difficulty of convincing companies to give up or share anything that gives them a competitive edge.
“We need consistency and interoperability at multiple places along that transactional journey,” he says.
The need for this will grow stronger as contact mobile phones are equipped with an NFC-like facility for contactless payments.
Data analytics in manufacturing, aviation and engineering
Crutchley agrees with Gallagher, that retailers are leading most current activity, and predicts some product suppliers will soon follow.
“Retail will move quickly, because retailers are very aggressive and brands see a new channel to get in touch with the consumer,” he says. “Until now, a brand would stick a product on a retailer’s shelf and its own experience from there is quite small. But, with the move to mobile, the brand can talk directly to the consumer.”
He says airports have also done some work, citing the example of Schiphol airport in Amsterdam, where the operator wanted to understand travellers' behaviour through arrivals, check-in, security, the retail area and boarding. There have also been indications of the banking and payments industry taking more interest.
Gallagher adds that there are signs of something similar happening with sensor data in engineering, providing the basis for predictive maintenance on vehicles or industrial plant. “There’s not massive take-off but I can see they’re beginning to use the same techniques,” he says.
It’s still a fractured picture, but Gallagher says the initial barriers have been overcome and that, as more tags are installed and more contactless payment cards issued, there will be scope to build a wide range of analytic technologies – what he describes as “a perfect storm” for innovation in the field.