Unearthing the customer: data mining is no longer the preserve of mathematical statisticians. Marketeers can also make a real, practical use of it - Revenue-Generating Networks

Data mining has an image problem -- an image as a dusty, theoretical discipline best left to mathematical statisticians burrowing through vast quantities of data in some dark corner of the office. It is grudgingly admitted to be useful, but still seen as light years away from the everyday work of marketing, operations and business managers.

However, all telcos aspire to provide better and mare profitable service to their customers and to attain operational excellence -- two business problems where data mining can be incredibly valuable. This is why data mining is beginning to shake off its 'strictly technical' tag and move into mainstream business use. The statisticians are still there, but only to provide support and develop new models for use by the business community.

Panning for gold

Data mining is an analytical tool that enables business executives to advance From describing historical customer behaviour to predicting the future. It enables companies to proactively manage business relationships, drive growth and answer complex questions. These questions might be: 'Who are your most profitable customers?' or 'How can you increase your levels of customer satisfaction, loyalty and lifetime value?' The findings can be used to identify business opportunities and to implement strategies that increase revenue, reduce expenses and offer new competitive advantages.

Mining the right seam

Data mining does not simply answer questions at the touch of a button. Rather, it is a multi-step process that includes defining a business problem, exploring and conditioning data, developing the model and deploying the knowledge gained. Typically, companies spend the bulk of their time pre-processing and conditioning data to make sure it is clean, consistent, and combined properly to produce business intelligence they can rely on. Moreover, successful data mining requires input that accurately reflects the entire business.

In the telecom industry, operators must understand where the power of data mining lies when tackling specific business challenges that may be either predicative and/or descriptive in nature. These might include:

* segmenting customers;

* predicting customer propensity to buy (or to churn);

* detecting fraud; and

* increasing organisational efficiency.

Once companies apply the data mining process correctly, they witness real results. For instance, one European mobile operator calculates revenues and costs for each customer, which means that it knows the actual value of each subscriber, not just the ARPU.

Data mining can also bring dividends in managing network performance. A leading US operator uses it to ensure that calls are routed effectively. This is done by continuous monitoring of performance rules and the analysis of data, both data on the history of component and trunk usage, and on the current network activity metrics. This operator has seen 'false' service and engineering call-outs decrease and the number of successful calls on their network increase Customer satisfaction is not the only plus. The information also enables the provide the quality and availability of service that the regulator demands.

Drilling deep

Many people believe you need an expensive, dedicated database, data mart or analytic server to mine data. These data marts are not only costly to purchase and maintain, but also require data extraction for each separate data-mining project, which is a major time-wasting process.

However, advances in database technology mean that data mining can now be done in an enterprise-wide data warehouse, which can also function as a customer and operational database. This means total cost of investment is considerably lowered.

By incorporating data mining extensions within a tdata warehouse, companies can reduce costs in other ways too. First, there is no need to purchase and maintain additional hardware dedicated solely to data mining. Secondly, companies minimise the need to move data in and out of the warehouse for data mining projects, which, as already noted, is a labour and resource-intensive process.

One of the largest mobile operators in the US uses a centralised EDW (enterprise data warehouse) to provide information on a wide variety of business applications from customer care to marketing. Previously, this operator had many different regional data marts, which made getting a holistic view very difficult. By transferring to a centralised EDW, this operator could get consistent information almost 90 per cent faster than through the previous fragmented data mart approach. This means the operator knows the decisions it makes tare based on the actual behaviour of their customers, rather than on gut instinct.

Pick and shovel work

In the early 90s, eighty-five per cent of the effort in data mining was in prepping the data and transferring it from its home database. Today, automated tools have cut data preparation and extraction to just 15 per cent of the work, and they have given normal businesses access to data mining -- often transforming them in the process by revamping management of, say, customer relation. The low level of preparation also leaves valuable time to discover completely new patterns in data.

All of this means that, though data mining can itself bring clarity to great complexity, it is not as arcane as its reputation. In practice, data mining is a collaborative effort bringing together knowledgeable personnel from three key areas. Business people must guide the project by creating a set of specific business questions and interpreting the patterns that emerge. Analytic modellers, with an understanding of data mining techniques, statistics, and tools must build a reliable model that addresses the questions asked. IT personnel provide insight into processing and understanding of the data, as well as supplying, key technical support.

In the telecom industry data mining is proven, delivering results and fast ROI (return on investment) in many areas including customer management, marketing campaign management, customer value measurement, financial management, revenue assurance and network performance optimisation.

All nuggets are valuable

The amount of customer data a company possesses is not an issue when it comes to deciding whether data mining is worthwhile. Any company with data that accurately reflects the business or its customers can build models that provide insight into important business challenges.

A mid-sized mobile operator in Asia-Pacific, for example, had a fraud problem. The company used a centralised database to increase the speed at which it could analyse CDRs (call retail records) and better understand the behaviour of its customer base. This enabled it to detect subscription and usage fraud within days rather than months, bringing it a sixty per cent reduction in account fraud loss.

At a time when operators are under increasing pressure, this ability to add significant contribution to the bottom line, without new subscribers or increased ARPU, had a major impact on the financial performance of this operator. Although the database had been purchased to specifically deal with the fraud problem, the operator found that it also provided many other benefits. By leveraging data from the data warehouse, savings across IT operations and system integration could be made allowing, in turn, for shorter application development cycles and a 30-50 per cent cost reduction as a consequence. Once these gains had been recognised within the company, other departments began to use the database -- giving the business as a whole fast access to critical information. This led to rapid decision making on areas such as churn, daily activations and revenue streams. Users found that information that was previously taking months or weeks to access was now available on the day.

As all of this shows, data mining is no longer too slow, or expensive, or complicated to work effectively. Telcos of all sizes are proving that data mining is essential in the search for gold in today's hotly competitive, customer-focused business world.

Martin Morgan, marketing & strategy director, communications industry, Teradata

(www.teradata.com)

COPYRIGHT 2003 Horizon House Publications, Inc.
COPYRIGHT 2003 Gale Group