The Data Grab

Posted 10/15/2000

The Data Grab
Customer Analysis Could Separate Winners From Losers
By Paula Bernier

E-commerce companies have become very sophisticated in their ability to reach the customer.

In a highly publicized case earlier this year, DoubleClick Inc. (www.doubleclick.com) had a plan to sell to other businesses a record of the web pages individual users had visited. Its intention was to compile that information by linking its own database of individuals' website trails with external databases including individuals' names, addresses and demographic details.

While that particular plan was nipped in the bud by privacy advocates, it points to the larger trend of web companies using customer intelligence to better identify their target customers.

Meanwhile, integrated service providers--which, unlike most e-tailers, are often armed with significant network expertise and financial assets, and may already hold massive amounts of customer data in their networks--tend to be much less sophisticated in gathering, analyzing and using customer data to retain existing customers and obtain potential ones.

Telcos are doing analysis, but they're not applying it in a very sophisticated fashion, says Christine Wright, vice president of CRM (customer relationship management) services at Convergys (www.convergys.com). Unlike e-tailers that drill down to the individual level, telcos tend to look at user groups, she says. But even analyzing customers at that level seems to have been a far tougher challenge than most telcos originally anticipated.

For years, telcos have been talking about plans to store customer data in data warehouses to be extracted for data mining. But because of the complexity involved, data analysis by service providers has been quite limited to date. The exception has typically been in high- churn product segments such as cellular and long distance.

According to Daniel Kenyon, general manager for communications industry strategy at PeopleSoft Inc. (www.peoplesoft.com), many data mining efforts that telcos did undertake became "black holes." PeopleSoft recently hired someone who worked on a data warehouse initiative at SBC Communications Inc. (www.sbc.com) for two years, and that effort ended up being scrapped because it was too cumbersome, says Kenyon. "We're trying to create data marts and offer prebuilt workbench capabilities so that they can do this stuff in weeks vs. years." he says.

While veteran service providers that have plenty of customer data within their networks may be weighed down by the very information that would seem like marketing gold, the new service providers haven't had the time to amass a significant customer base and the related customer data, says Michael Keeler, president and CEO of DataZen Corp. (www.datazen.com), a data mining and analysis company that aims to become the leader in customer acquisition.

But to succeed, service providers are going to have to find a way around these many barriers to better analyze network and customer data to reduce churn, increase customer satisfaction, and otherwise allow for more efficient and profitable businesses.

"I think this will determine who's going to win vs. who dies," says Keeler.

CRM

A key in this movement is CRM.

Kenyon of PeopleSoft, which brought CRM into its product mix with the recent acquisition of Vantive, describes CRM as an approach to how companies work with their customers and how they manipulate their software. "It's an approach to setting up a solution that centers around a point of the customer service representation," says Kenyon. "You can gain a view using our software to the entire relationship the customer has with the company--whether it's a phone call, a truck roll, whether they have to send you a replacement part, whether they have to ship another modem for DSL."

What Kenyon is describing is the screen that the customer service representative (CSR) at the call center sees when interacting with a particular customer. But once entered in the system, that information can then be shared among various departments in a service provider. The sales force, the field force and management, for example, can then analyze that customer information to see if particular customer or product segments are profitable, and how to better target products to particular customers or customer groups. PeopleSoft's products also tie into a company's billing, supply chain and other systems to enable the organization to get a broader view of customer behavior and service, such as what services the customer ordered or how many visits it took for a field rep to install a particular DSL connection.

Rich Aroian, vice president of marketing for the ADC Software Systems Division of ADC Telecommunications Inc. (www.adc.com), adds that while what a customer ordered is very valuable information, what may be even more important is what a customer chose not to buy. "Many times any kind of dialogue you have with a client, even if it doesn't result in inputting data about service, could [be used] for data mining," he says. Service providers that can capture that kind of information--whether through a sales call, call center or website--may have better success at selling that service or others like it in the future.

"You buy CRM to help customers and to help you be more profitable," says Kenyon.

Sprint Corp. (www.sprint.com) is the first service provider to use PeopleSoft's CRM For Communications product. It uses the tool to process orders and analyze the success of specific marketing campaigns. The product also enables Sprint to track a variety of customer attributes, such as whether the customer is a business or consumer, part of a family or an individual, and whether the customer purchases additional services from other network operators, for example.

PeopleSoft recently released enterprise performance management (EPM) functionality for its CRM product. Kenyon explains that EPM enables data to be imported from other sources, such as internal or external databases, different pieces of an operational support system or virtually any other source. That data can then be compiled in a data warehouse. And the PeopleSoft process can be used to extract such data from various data marts (that is, mini-data warehouses) and put them into workbenches to do various types of analysis.

"CRM really doesn't address the needs of the corporation in terms of looking at the population of your customer or prospects. CRM instead creates relationships with the customer and the company itself," Kenyon explains. "EPM tracks the relationship over a given population. CRM allows you to track the information necessary to make those promotional decisions, but EPM allows you to view that information."

Convergys also offers a variety of products in the CRM space. Wright explains that the company has a data warehouse that enables its customers to import data from customer contact centers and billing systems. Three toolsets sit on top of that.

The SAS Institute Inc.'s (www.sas.com) Enterprise Miner is a statistical tool used principally to determine relationships between variables like key characteristics of customers' usage patterns and their propensity to churn. It looks at customers' ages, incomes, where they live, whether it's an urban or rural area, and their propensity to be upsold or cross-sold. It also looks at the optimal rate plan for that customer based on his or her usage. It then scores the customer based on his or her ranking in each of those categories and assigns the customer to a particular segment.

Recognition Systems' (www.protagona.com) Protagona is a campaign management toolset that splits customers into segments, establishes different types of contacts, sets procedures on how to treat customers differently depending on their attributes, and assigns a campaign for the group into which that customer fits. For example, says Wright, a carrier might want to route, queue or tailor content differently based on a customer's value, propensity to purchase additional products or whatever. The Convergys Business Rules Decisioning Engine, a real-time database at customer contact centers, meanwhile, could direct higher value customers to a special customer service rep while lower value group customers are directed to an automated voice response system.

The third toolset is called Business Objects, and it does online analytical programming for management reporting and back-end tracking. For example, if a service provider wanted to compare all the customers in its high value segment with this month's revenue and sales report, it could go in and create reports on the fly to cross-section various pieces of data. Or it could compare the impact of a particular marketing campaign.

Wright says new real-time analysis developments will allow for even more sophisticated data manipulation. For example, the newest generation of GMAX, a programming language based on genetic programming, can analyze data in real time and learn from that data, so there's a continuous stream of data and learning. This technology is now being used primarily for web-server interaction, but some telcos such as Sprint are starting to look at it for such applications as real-time use of data for loyalty programs for high-value customers.

But today, churn remains the main application for data mining and analysis.

"Most customers are using [data analysis] for churn [prevention]," Jeff Caplan, vice president of solutions marketing of Verbind Inc. (www.verbind.com), which is helping service providers to stem churn and develop and assess marketing programs using a service bureau model. "Even though it's not exciting, it's a significant business issue." Verbind looks at individual behavior, not just group trending. Caplan says businesses often outgrow their services and don't necessarily learn about new options from their existing carriers, so they may reevaluate their carrier as they reevaluate their services. "This is an opportunity to upsell although it's primarily a churn issue," says Caplan.

"We not only say who is at risk, we also say why they are at risk," he adds.

Competing systems simply score customers based on their perceived potential to churn, he says. They also tend to look at customer behavior by comparing it to the past behavior of customer groups rather than identifying that a particular customer's behavior has changed. By looking at the individual customer, a service provider can take steps--such as spending more time on the phone line with a customer in an attempt to upsell that customer--to keep the customer, says Caplan.

"We want our clients to be able to return to an age in which they have a better, more personal relationship with their customers," he says.

Barry Shaw, director of BSS software development at Sigma Systems Group Canada Inc. (www.sigma-systems.com), which sells a product called CLEC in a Box for customer care and billing, says data mining should even be extended outside the service provider to the customers themselves.

"We're trying to take data mining as far as we can into GIS [geographic information services] systems and graphical analyses relevant to the client information and helping the client utilize the info," he says. "There are two points of view, one from the telco's point of view, the other is to help the client understand his own utilization of service."

Early Analysis

But data analysis on customers and the networks that serve them can and should start even before customers are acquired and services are introduced.

For example, prequalifying customers for services like DSL can be a deciding factor in whether a service provider succeeds or fails in the marketplace, says DataZen's Keeler.

According to Keeler, 70 percent of orders that come in fall through the cracks due to order entry/order management problems such as the ILEC kicking back an incomplete request, or because customers are outside the carrier's footprint, blocked from service because the copper in their area is not suited for DSL or are too far away from the switching office.

Telcos are facing huge turnovers in their sales staffs because the sales people will go out and close a deal, but the telco may only be able to provision service for half of those orders, so the salespeople don't get paid for all their work, says Keeler.

"It's really a time to revenue issue as well as an ROI [return on investment] issue," he says.

DataZen's ServiceFinder allows the service provider on its website to enable the customer to request service and automatically get a response to see whether service can be made available to that customer.

"We have developed an algorithm that works from the CO, and it traverses the street network to create a polygon that is a certain distance away from the central office," he says. It gives the route, the length of route and defines the wires that most likely would carry that user's traffic. And it makes sure that potential customer is within range of the CO for DSL, for example, and checks whether network resources, such as ports on the DSLAM, are available to that customer.

DataZen also provides a variety of other products and services relating to data acquisition and analysis. It has tools for market segmentation and analysis.

A carrier could use these tools to identify companies with 20 to 100 employees and $3 million to $5 million in revenues to ensure it builds its network to reach its target audience.

DataZen also operates a portal through which companies can buy data about markets, buildings, companies, decision makers in those companies and existing infrastructure in place in various markets by other providers. Its "intelligence factory" behind the portal can take data from those public sources and integrate that with customer data a carrier already has. That way a carrier could see where all CO facilities are in a given area, for example, so the service provider could do segmentation analysis and point salespeople only to potential customers that are prequalified for their services.

Choice One Communications Inc. (www.choiceone.com) and Teligent Inc. (www.teligent.com) are DataZen customers.

Preparing to roll out its high-speed cable-based Internet service, Australian service provider Telstra Corp. Ltd. (www.telstra.com) used Hewlett-Packard Co.'s (www.hp.com) Smart Internet Usage Program software to gather and analyze usage data to help decide what the best pricing tiers would be for the service, says Dana Kreitter, marketing manager for that HP product.

The other thing important to Telstra was to look at usage data to define its local content caching strategy. To optimize content locally for quicker delivery and more efficient use of its network, Telstra metered the usage of what content people were accessing most and analyzed that data to figure out what content should be cached locally and where. Kreitter says Telstra was a pioneer in analyzing content demand and caching in response to that information.

Kreitter believes service provider data analysis will increase as more IP services are added to the network.

"I think that's going to change with Internet services because with phone service you have a great mechanism for collecting the information [via call detail records from CLASS switches], but it's a limited range of services. But with the Internet, you have everything from telephony to commerce for anything. There is technology current or evolving that will make it possible for service providers to know what users' purchasing preferences are as well as how they use the network."

Share this article: Email, Slashdot, Digg, Del.icio.us, Yahoo!MyWeb, Windows Live Favorites, Furl
Add this article feed to: RSS, My Yahoo, Newsgator, Bloglines