After 18 months that have changed the Banking and Finance world forever, what was deemed an acceptable risk position this time last year is today being viewed in a different light. Banking groups are looking to lessen their risk exposure while still maintaining their share of the customer’s wallet. But with capital expenditure unavailable, they are turning to automation of previously manual activities to achieve their goals.

The latest figures from APACS show that fraud losses increased significantly in 2008 despite the impact of chip and pin. Card fraud overall rose by 14% last year to £609.9m while card-not-present fraud in transactions conducted over the phone, online or by mail order was up 13%, accounting for more than half of all card fraud. And while counterfeit fraud losses climbed by 18%, APACS says much of this was down to fraudsters cloning UK cards for use in European countries where chip and pin has not yet been rolled out.

Chip and pin may be effectively curbing the ability of criminals to operate on the UK high street, but organised crime has found and will continue to find new ways to operate. On-line fraud is growing too, with the use of trojans to hijack computers and keylogging to steal passwords and account details. The payments association found that on-line fraud had more than doubled to £52.2m in 2008 from £22.6m the year before.

What banks need to do is more effectively stem fraud losses at source. In the past, this has been done through the contact centre, traditionally the focal point for much of the banks’ interactions with customers. But today, many are reviewing operating costs in their contact centres, some 80% of which are tied up in the cost of agents themselves. They’ve toyed with the idea of offshoring contact centres to lower-cost locations (and employ a host of other options for keeping costs down, from home agents to IVR and speech recognition) but some are now bringing them back onshore as they suffer a customer backlash.

So with traditional contact centres in many cases proving an unacceptably costly overhead, many banks are turning to virtual agents and SMS contact strategies to help mitigate fraud risk, using Auto-resolution™ technology that for example automatically dials or texts customers to check on potentially fraudulent transactions.

Co-op looks to automate

When Co-operative Financial Services was facing up to rising fraud last year, it decided that the significant expansion that would be necessary in its contact centre just wasn’t an option. Instead it turned to the Auto-resolution technology it was already using in its debt collections business to tackle fraud. CFS, which includes the Co-operative Bank and internet bank smile, knew the technology could do a job for it in fraud reduction – and because the system it chose is securely hosted externally, it avoided many of the issues around integration and testing.

The likelihood of fraud is uncovered by FICO’s Falcon Fraud detection monitoring systems with complex algorythms checking for suspicious-looking transactions. What Auto-resolution allows banks to do is follow up more of these transactions and get through to more customers without having to hire dozens of extra fraud analysts.

Because the follow-up is a completely automated process using pre-recorded scripts and natural speech technology to interact with the card holder, it can run 24/7, validating their identity, and providing them with the opportunity to confirm any number of transactions. Martin Kent, Financial Crime MI and Analysis manager at CFS said: “CFS now makes around 12,000 of these automated calls every month, most of which are new calls that we weren’t previously making.”

Customers don’t mind these calls – in fact they welcome anything the bank is doing on their behalf to try to combat fraud. One CFS customer wrote: “Not a minute after I finished the online bookings, the telephone rang and it was the CFS Fraud Prevention system. I was impressed at how efficient and easy to use this was. A great innovation.”

Once the Fraud detection system classifies the type of transaction – for example internet, ATM, card-not-present or general – this information is relayed to the customer to provide as much information as possible so they can recognise genuine transactions. If the customer cannot verify their identity, or the transactions, they are immediately connected to a fraud analyst who will investigate further.

The results at CFS are impressive too, with unrecoverable card fraud losses having fallen by 70% since the implementation, loss per case down by 50% and overall card fraud losses quarter on quarter down by 35%. When fraud is discovered, the key is to get to the cardholder as quickly as possible and if necessary stop the card; hence the speed of the automated system helps deliver these impressive results.

What banks are finding, prior to using automated systems, is that their agents are spending too much of their time trying to get hold of customers, leaving messages on voicemail or chasing up different telephone numbers. Not only is this frustrating for the agents it also means that while the right person is being found, the clock is ticking and further fraudulent transactions could be racked up.

The automated system can still get voicemail over 50% of the time, and makes contact with the right person around 25% of the time, but as they can dial a potentially limitless number of times, this no longer matters. The system also frees fraud analysts to work on the higher-probability cases, and as soon as the system discovers a fraudulent transaction, they can take over.

If the potentially fraudulent transaction is not validated quickly, and fraud losses start mounting up, the alternative is to stop the card. But every card that is mistakenly blocked means lost revenues for the issuer in transaction fees, additional costs to the bank to unblock the card and an inconvenienced and dissatisfied customer. When the card is subsequently reactivated, the chances are it will have gone to the back of the cardholder’s wallet, and will have less chance of being used.

Secrets of success

The use of virtual agent technology in preventing fraud is somewhat counterintuitive in the banking sector. After all, there’s a perception that given a choice, customers will prefer to talk to a human being than a machine. So what makes this application of the technology so successful?

First, it’s picking the right use of the technology – and fraud investigations, automating collections activity and marketing activities such as the activation of new or replacement cards – are a perfect fit for virtual agents. Customers might not welcome talking to a machine in certain situations, but when they understand the need to confirm a transaction was theirs to cut out fraud, or when they are having what could potentially be a sensitive conversation about collections, the technology is invariably welcomed.

The key is to communicate with customers so they know what is coming and are not frightened off. Best practice has uncovered a range of methods for communicating a new automated service – with a screen or flash on the ATM machine, posters and leaflets in branches, holding pages or advertising on the bank’s website, information given out over the phone during the hold music, and statement messages and inserts.

One of the few barriers to the success of such an application is consumers’ suspicion that the call might not be a legitimate one, so banks work hard on the sound of the voice, the script used and the information given out to reassure the customer. Customers should be told how they can recognise that the call is from the bank and not a phoney call – for example, telling the customer to expect personal information to be given and for the system to only ask for verification of partial information.

The second key to success is to provide an end-to-end real-time solution with the ability to duck out and speak to a human agent at any point. Some technology providers will only provide a partial solution or will not have had experience integrating with transaction monitoring systems – and that can lead to problems. With fully integrated and automated systems, transactions are passed over to Auto-resolution in real-time, and that in turn updates the case management system with any communications that have taken place.

If the customer does feel the need to duck out of the automated system and speak to an agent – if they are suspicious that the call is not genuine or want to discuss things further – they will not need to run through the whole identification process again, because agents will be able to pick up where the system left off.

A third critical factor for success is to keep on tweaking the system for optimum effectiveness. The great thing about automated self-service systems is that they provide so much management information back about their effectiveness. Fraud managers can monitor at exactly what point customers are dropping out of calls – for example if a particular phrase in the script scares them off. They can monitor the number of calls being made and the amount of traffic being generated. And obviously, they can quickly and easily see how much fraud has been identified and caught.

Analysis of these numbers tells us, for example, the importance of the first call about a potentially fraudulent transaction. According to data from one bank’s real-time credit fraud system, it is three times as likely to make contact with the consumer on the first calling attempt, the consumer is three times less likely to disconnect during the first calling attempt and is more likely to confirm a fraud on the first calling attempt, hence the need to get to the consumer quickly and efficiently to rule out fraud.

The acceptance of virtual agents and automated customer contact is evidenced by customers reporting high levels of success for their applications. Consistently when banks and financial organisations survey their customers after an implementation, they indicate very high approval ratings for the service. One UK bank found customers preferred its virtual agents because they were non-judgmental and they felt in control of the call. Another bank, this time based in France, found that customers would be willing to pay an additional euro to keep the service – so it could even be positioned as a premium offering.

While banks are adopting technology and self-service everywhere they can to improve efficiency – from internet banking to ATMs and kiosks – using it to combat fraud has to be a priority. Nine of the top ten UK card issuers and eight of North America’s top ten financial institutions are using auto-resolution services, so it presents a perfect opportunity to boost collections, enhance your marketing campaigns and attack fraud where it really counts.