Ending the churn trail


The concept of networks working to clamp down on the customer churn merry-go-round is not a new one, with all investing significant resources into this area over the past four years. But the market is now tighter than ever, thanks to its saturation and the onset of the recession; networks must work harder than ever to hold onto their customer base and attempt to draw numbers from their rivals.

Further investments are being made into customer profiling, to paint an increasingly detailed picture of customer segments and individual customers. But there is a divide over what approach is more effective to prevent churn – using customer profiling to enhance customer loyalty through targeted campaigns, or using a detailed analysis of contact groups to identify likely churners and intercepting them before they move.

All of the five UK mobile networks but 3 are said to be implementing social network analysis (SNA), which uses a series of complex algorithms to track and map customers’ calling circles and patterns to formulate “churn pressure” scores and identify whether a customer is likely to churn in advance of them doing so.

Dublin-based software company Idiro, used by O2 UK, offers such technology based on  the theory that churn is “viral” – that if some of a particular person’s close contacts move networks, that person is likely to follow. Idiro director of sales and marketing Simon Rees claims the company builds maps of “who is connected to who” and runs its algorithms to identify who is most likely to churn next.

“About a quarter of all churn is contagious. Churn has the biggest impact in terms of monetary value to a network and has a bigger sense of urgency now in terms of the economic situation,” he explains.

“All marketers want to understand how social relationships work and who has influence over decisions. All our decisions are contagious – it’s part of group politics. You can use records of who has made what calls to who to analyse communities and understand relationships between users.”

Networking analysis

SNA uses the philosophy that people who receive a lot of calls typically have more influence within a calling group, identifiable through call pattern maps to show who is calling who, how long the call is, the time of the call and which day it is made. Idiro can also generate a “churn pressure score” to indicate how much pressure a customer is under to churn due to the fact other people in their calling community have recently churned. There is also a “churn influence score” to qualify how many people a given subscriber is likely to take with them if they do churn.

“The better an operator can identify valuable people the more return it will make,” Rees says. “They might spend £200 saving a customer from churning but they might also have saved this person’s best contact, for example their sister, for free.”

Idiro claims it is able to predict churners who might otherwise not be obvious to the network, with one of its case studies boasting an average 1,500 extra churners predicted accurately each month. This is a 30 per cent increase on top of the network’s own in-house churn analysis activity, which Rees argues can be basic and inconsistent. He claims that O2’s dominance of the UK market is down to its clever use of analytics. “It is no accident that O2 is doing so well in the UK.”

Rival company Infosys, based in Bangalore, India, has yet to snare a contract with a UK network but is currently in negotiations for this. While it can also offer customer profiling and churn prediction based on SNA, it comes from a different standpoint to Idiro. Networks can use information from both companies to formulate marketing campaigns, but Infosys’ business model is based on identifying high-value customers and devising strategies to bolster their lifetime value through its “customer value management” platform.

Infosys associate vice president Deepak Swamy reckons that the company’s research has found networks haven’t been putting data to good enough use when it comes to driving marketing programmes to engage customers over a lifetime.

“There seemed to be a gap in tying marketing campaigns into real ongoing engagement with the customer. We create a ‘customer value management’ platform which allows a network to use data from different sources to create ongoing engagement with an end user, such as what handset they use, when they last upgraded and for what reason,” Swamy explains.

“It can all be tied together in terms of creating offers and delivering to various market segments and understanding the performance of a particular campaign to see how an entire segment will respond.”

Infosys works on the hypothesis that, in a similar fashion to the financial services industry, around five to 10 per cent of a mobile network’s customer base contributes 70 per cent of its profits. The vast “middle” base of customers is marginally profitable, and there is a bottom segment that is considered unprofitable.

Put simply, Infosys’ message is that rewarding high spenders is the key, and customer profiles can be used to create tailored offers. “Clearly, all customers aren’t equal so operators should have mechanisms to recognise their important customers,” Swamy argues. “Credit card companies recognise premium customers with special offers and this is not necessarily so in the mobile industry – premium customers don’t get treated dramatically differently than regular customers. Premium customers want to be acknowledged so the operator should be rewarding them for their business.”

Beyond the SNA-based calling maps, there is a wealth of information that can be legally obtained by a network about its customers. Infosys’ processes analyse the customer’s actual interactions with the network, when and why they have contacted customer services, what services a customer subscribes to, and whether they are a heavy data user. A customer’s age can also be used to predict how their usage might evolve over the next five years.

Information can also be obtained from third party sources, such as the credit check performed when a customer first joins the network, which can give a basic financial profile of a customer. Networks can also purchase market segmentation information from data syndication agencies.

Based on a postcode, syndicated data providers can give a profile of the residents that live in that area – what kind of houses they live in, what cars they drive, where they might shop. 


For full article, see Mobile News issue 444. To subscribe, click here