Telecom Data Mining: It’s Time to See Its Value

By Michael Haggerty, Vertek

Telecommunications is one of the most data-intensive industries in the world, and a great opportunity exists for telecom managers to analyze the large amounts of data that have been collected in their network databases in order to improve the short-term and long-term operations of their organizations. One highly effective tool to aid in this data analysis is the proven process of data mining.

Data mining has been used for years to analyze data in two or more fields within a relational database. Typically, specialized software is needed and doesn’t have to be costly. Moderately priced software is available that can get almost anyone started. Data mining enables the user to view data from many different perspectives, categorize the data in new ways and summarize the resulting relationships between seemingly incongruous pieces of data that the software has identified. Careful analysis of these relationships can provide managers with the ability to optimize internal network operations and better manage external customer-facing activities such as churn and marketing.

Ironically, with such an enviable wealth and diversity of data at their fingertips, many telecom managers have been reluctant to use data mining to their advantage. Arguments for this can run the gamut from too expensive to not enough time to lack of upper-management commitment. However, raising arguments like these will result, quite simply, in lost opportunities for many telcos. Conversely, if data mining is undertaken in a controlled fashion, many new opportunities will surface that will enable a company to become more profitable and competitive. And after all, the data already exists, and managers need to make better use of it.

An important point to remember when considering data mining is that first and foremost, data mining is always a business activity. In order to arrive at meaningful results, a carrier needs to align data mining with the goals and objectives of its business. Otherwise the results will be irrelevant and lacking context.  Data mining should also be focused on exploring different hypotheses, taking into account disparate data, where the end result of the data mining exercise could potentially drive building a better customer experience, creating new operational processes and improving the overall bottom line.

Telecom data about customers such as call detail and customer information can be profitably data mined. Mining these data types can help you determine customer behavior and identify opportunities to support the goals of expansion of your customer base and reduction of customer churn. 

Customer churn is an area worth mining, as it is becoming ever more important to retain customers and improve wallet share. Mining churn rates tied to the number of trouble tickets issued in a 12-month period might uncover a correlation between the two. Perhaps after three trouble tickets, the customer leaves the network, rather than after two tickets. This analysis might prompt the carrier to flag all customers with two trouble tickets and inaugurate some action whose purpose is to retain those customers. Following those remedial actions the carrier should mine once again, and determine how many customers at possible risk remained with the carrier.

Data mining can also be invaluable in the development of marketing programs. For instance, a carrier might have a goal to increase the number of subscribers who pay their bills online, thereby reducing the cost of paper, printing, postage and handling. Mining the carrier’s database of online bill paying subscribers and the database of active users purchasing content on the telco portal might determine that a low percentage of subscribers using the content portal pay their bills online. A decision could then be made to place a link on a content portal page that says “pay your bill here." If an upswing in online payments occurs, the carrier could mine again and determine the amount that billing costs were reduced, if at all, and whether the goal of decreased billing costs was achieved.

If a carrier has a business goal of increased revenue from advertisers, data mining can be used to analyze a carrier’s data on portal usage by time of day. If usage is consistently up one or two times a day, every day, then the carrier might want to consider charging advertisers higher fees for the privilege of advertising during those times.

And let’s not forget about all the data mining that can be done around networks and equipment. Data mining of mean time between failures (MTBF) might result in a correlation between MTBF and other pieces of CPE. Vendor servers might be deployed in a network, and over a period of time it is determined that 30 percent of these servers consistently crash after 12.5 months.  The decision can then be made to either replace the vendor servers every 12.2 months or find more reliable servers from another vendor.

Above all else, once a carrier begins data mining, the tool should never be abandoned. Data mining should consistently be used as a means to proactively achieve the carrier’s business goals and objectives. There is a wealth of information in every carrier’s database just waiting to be mined, and a wealth of new ways to improve every carrier’s business as a result of data mining. The data is there. It’s now up to every carrier to have the forethought to analyze it, not think of it as a resource only after business goals are not met and revenue or a competitive advantage has been lost.

Michael Haggerty is director of business assurance at Vertek Corp. He has over 8 years of Telecommunication industry experience, specifically in the core area of Financial and Revenue Assurance. Mike as lead over 40 complex Revenue Assurance audits for Vertek, which required an in-depth knowledge of billing systems and processes. His expertise in the Financial Assurance area includes Project Management, Process Optimization, Business Requirements and Operational Management. He can be reached at MHaggerty@Vertek.com.