CRM, one step at a time, by CRM guru

MyCustomer.com
Share this content

Deploying CRM analytics companywide via a data warehouse is far too complex and costly for companies to do all at once. Fortunately, there are ways companies can simplify the process. Many are focusing CRM analytics projects on areas where they have the closest contact with customers: website transactions, suggestion lines, and corporate call centers.

CRM analytics expands on traditional data mining by applying statistical and reporting tools to information culled from these more personalized customer contacts.

In the past, analytics have been based on historical analyses of customer behavior and demographics. But the new model is to run the analysis closer to the time when the data is captured – ideally in real time.

The right offer
“The faster you can collect and analyze data, the better position you’re in to make the right offer to the right customer at the right time,” said Greg Gillett, national director of Cap Gemini Ernst & Young’s CRM real-time marketing practice.

Armed with reports generated from CRM analytics, companies aim to more accurately predict customer behavior, identify buying trends and ferreting out the reasons for their marketing successes or failures.

Revealing defects
An example of a company that’s using this approach is eMachines Inc., Irvine, California. The computer maker uses Alorica Inc’s Helix software to learn more about its call-center operation. Using Helix, eMachines has built a database of customer problems and comments, as well as call-center actions. The software aggregates the information and produces reports that may reveal a defective part, such as a power supply, on a particular model.

If the model is still in production, the company can solve the problem before more units are shipped. Or the report may indicate specific problems users are having with the latest version of an operating system.

An increase in calls about configuring Windows Millennium, for example, may indicate a need to beef up training in that area for call-center staff. And if the analysis indicates that customers are waiting too long on hold and are hanging up in large numbers, the staffing schedule may need to be adjusted.

Hung up
EMachines COO Jack Ferry said when implementing Helix, he had to make decisions on business rules and categories. For example, Ferry decided that the system shouldn’t include in its metrics customers who hang up within the first 30 seconds. He said most customers in that group hang up because they change their minds about getting support at that time, not because they are frustrated by a long wait.

Another way companies use CRM analytics is to capture a broader range of corporate data. One example is Hilton Hotels Corp., which uses E.piphany E.4 at its Beverly Hills, California, headquarters to analyze data generated by its reservation and property management systems, as well as from loyalty programs such as the Hilton Honors program database. The Honors database contains information about customers most likely to use Hilton Hotels. And since members of the program provide Hilton with personal information, the database also contains valuable demographic information.

Hilton marketing employees use the analysis performed on the information in those databases to create direct-mail campaigns. The analysis also helps hotel managers plan for seasonal activity. For example, it lets them determine how many rooms to reserve for wholesale customers and how many for business travelers.

Reports in 30 minutes
The reports aren’t new to Hilton managers. In the past, they could request the reports from IT. But the checkout line was usually three to six weeks long. “By the time they’d get the report, it was often too late to act on it,” said Joanne Flinn, vice president of leisure marketing, who adds that users can usually receive reports in 30 minutes or less.

Fingerhut
Despite the availability of packaged CRM analytics products, some companies need the versatility of a data warehouse. Fingerhut Companies Inc., Minnetonka, Minnesota, which sells electronics, housewares, jewelry, apparel, flowers, and speciality food gifts through catalogs, is an old-line database marketer that depends heavily on data analysis from its fulfilment and distribution systems. By segmenting and analyzing its customer base, Fingerhut can conduct more timely mail-order campaigns.

Randy Erdahl, the company’s director of business intelligence, says Fingerhut has done a good job of understanding its customers. But as it moves into e-commerce, the company has to prove itself all over again.

Erdahl said that although Fingerhut started building its data warehouse in 1995, it isn’t ready for the data the company is getting from its website. Fingerhut needed to establish new categories, including such measures as how much time customers spend on a page and which pages tend to lead customers to make a purchase.

So, Erdahl is using homegrown extraction, transformation and loading (ETL) tools to fetch data from the data warehouse as well as from web traffic logs. The data is stored in an operational data store. The company’s 30 analysts can then run queries against the data store. The company is also augmenting its set of analytics tools to support eCommerce.

“With eCommerce, we have to act much more quickly than with catalog sales,” Erdahl said. “So we’ve brought in tools that can help non-statisticians make marketing decisions.”

Quadstone
Besides the statistical software from SAS and SPSS, Fingerhut uses the Quadstone System, Quadstone’s CRM analytics software. The software includes built-in intelligence that can create subcategories based on customer age. It also creates an easy-to-understand graphical scorecard, which shows the value of each web page.

Erdahl hasn’t given up on augmenting the data warehouse to make it CRM-ready. The advantage of marrying the data among multiple channels is that it lets the company do more granular analysis, using categories and segments that may not have been preconceived and included in the ETL tools. Erdahl said Fingerhut will try to complete its data warehouse project this year.

CRM analytics
An important point about CRM analytics is that it’s the type of company, not the size, that determines the most appropriate kind of CRM analytics project. Even large corporations may need only a limited CRM project.

Pillsbury, Minneapolis, uses CRM analytics for only one type of customer interaction: calls received on its toll-free comment line. The company receives thousands of calls each year. It’s now developing a set of analytics tools that will let managers throughout the company analyze the data collected from those calls.

Slow delivery
Pillsbury never threw out all those comments. Still, reports based on the data were created slowly. Both monthly and quarterly, managers would receive printed analyses of the comments. The slow delivery hampered progress. For example, when numerous customers called in to say they’d love to see chocolate-flavored Toaster Strudel, it took the company months to even begin thinking about developing the new product.

“We want to be in a position to react much more quickly to what our customers want,” said Fred Hulting, a senior research scientist at Pillsbury. So, Hulting and others at Pillsbury built a new system using Insightful Corp.’s S-Plus analytics tool and its StatServer product, which lets web users perform analyses on S-Plus data.

Hulting said deployment was time consuming. He and his staff had to decide on which types of analyses business managers would want to run. He then had to create a complete specification for each analysis, which included S-Plus code and the various parameters required to define the inputs and outputs, as well as how the analytics and its results were displayed to users.

Some companies choose to farm out their analytics projects. For example, when British Airways wanted to increase the effectiveness of its website, it decided to outsource, said John Mornement, senior manager of eCommerce operations and design.

Mothballed
The airline hired WhiteCross Systems to analyze the two Gbytes of web traffic data the system collects, as well as to generate reports. Some of the analysis pointed directly to the fix. For example, dead links and pages that haven’t been accessed for months are taken off the site daily. With the first report, the airline was able to mothball hundreds of obsolete pages.

Companies looking to deploy CRM analytics have options ranging from a full-fledged data warehouse to targeted deployments and outsourcing. While outsourcing will likely expand, expect most companies to take a more targeted approach, one slice at a time.

Larry Stevens

Replies

Please login or register to join the discussion.

There are currently no replies, be the first to post a reply.