By GEOFF NAIRN(Published in the Financial Times on September 16, 2009. Read the original here.)
Extracting more value from existing customers is a priority for businesses in the current economic climate – and technology is helping to achieve it.
It uses complex statistical models and algorithms as well as other data mining and analysis tools.
In a recent Forrester Research survey of more than 350 CRM professionals, “improving customer loyalty” was rated as a key priority by 57 per cent of respondents and came second only to “improving the customer experience”.
“You cannot afford to lose any important customers but many companies do not have the processes to identify dissatisfied ones,” says Sam Israelit, a partner at Bain & Company.
This is where predictive analytics can help, by identifying early-warning signs that customers may be unhappy. For example, in the phone industry, where the problem of “customer churn” is pronounced, dissatisfaction becomes obvious when a customer starts forwarding calls to a different number.
“But if you wait for them to set up call-forwarding, it may be too late,” says Colin Shearer, senior vice-president of strategic analytics at SPSS, a leading vendor of data analysis software which recently agreed to be bought by IBM.
Cablecom, Switzerland’s largest cable operator, discovered unhappy customers were most likely to quit after
about nine months. So, it ran a feedback programme targeting people who had been customers for seven months.
The data from the programme was analysed using SPSS software and more than 100 churn indicators identified – their exact nature is a closely guarded secret – but they have enabled the company to identify customers who may be about to leave.
The “customer retention team” is then activated before it’s too late. In pilot studies, the technology has allowed Cablecom to reduce churn rates from 19 per cent to just 2 per cent.
Churn is also a problem in financial services. In 2005, Marks & Spencer Financial Services had to issue chip- based credit cards to its 2.5m customers in the UK. It was worried the disruption would reduce revenues and increase churn. So it used predictive analytics technology from a UK company, Kognitio, to identify the most profitable customers, who were given rewards for fast activation.
Roger Llewellyn, Kognitio chief executive, says the M&S example turned what would have been a “cost sink” –
sending out new cards – into an opportunity to reward customer loyalty and increase cross-selling.
Data analytics also allows businesses to identity customers who are unprofitable or only marginally profitable. Typically, up to 20 per cent are unprofitable but it can be higher – in the 1990s, about 60 per cent of household accounts were loss-making for US banks.
Technology can reduce the cost of serving lesser value customers, by making them use interactive voice
response systems rather than live agents, for example. Some businesses have even “fired” their worst customers. American Express, the credit card company, for example, “invited” a small group of high-risk cardholders to pay off their balance and close their accounts. It offered them $300 if they did so quickly.
But there are dangers: “You have to be sure you have accurately calculated the future profitability of those customers,” says Jeff Gilleland, product marketing manager with SAS, the US business intelligence company.
Other technologies also save businesses money on dealing with customers who expect to be able to speak to a representative if they have a query, complaint or want technical support. Such calls do not generate revenue directly and cost a lot to handle.
“The most expensive channel is when customers call us,” says Chuck Udzinski, consumer services manager at
Black & Decker, the US power tools maker.
Black & Decker now encourages customers to use its website for support rather than the call centre. The system uses web self-service technology from RightNow, a hosted CRM vendor that also supplies Black & Decker’s main CRM system. According to Mr Udzinski, web self-service saves the company $250,000 a year by deflecting tens of thousands of calls.
“The days of the contact centre being just voice are long gone,” says Paul Segre, chief executive of Genesys, the
US specialist in contact centre software.
But if not integrated correctly, low-cost channels can backfire.
Mr Segre gives the example of a customer applying for a mortgage online: if the customer hits problems during the process, the system should recognise this and pop up a window asking if they want to chat or be called by an agent.
“If you offer different ways to contact the business, you can dramatically increase completion rates,” he says. Predictive analytics is thus being used to analyse how customers use websites and to tie them more closely to
other communications channels.
“If I am an existing customer and I am looking at mortgage rates on a website then that intelligence could be harvested and I may receive a call from a mortgage broker,” says Mr Gilleland of SAS.
Predictive analytics can provide much-needed insight into customer behaviour and is no longer a technology reserved for the biggest banks and retail giants. Sofware-as-a-service offerings – Kognitio offers one – and new low-cost data warehouses from the likes of Netezza have put these tools within reach of a much wider market.