Data Mining: A Strategic Plan

Feb 26, 2008 2:19 AM  By

There is great advice out there about the how-to of data mining. But there is precious little about the why-do.

DMers often launch a campaign under tight deadlines and fail to think through how they’ll use customer data to support it or measure the results. In effect, they create more work and risk for themselves.

If only they focused on strategy as much as they do on tactics. All would be well.

Where to start?

First, the marketer should streamline the preparatory work for each campaign by building a consistent method for identifying opportunities within the customer base.

This builds efficiencies into the marketing process and helps deliver relevant and consistent customer experience across multiple channels, thereby improving marketing efficiency and effectiveness.

The tools for tactical data mining are widely available and generally reliable, making it easy to apply them to one-off efforts. But there is no software package to guide marketers through the strategic approach; they have to step back and do the thinking themselves or engage a partner who can help guide them through the process.

What follows are the six stages in the hierarchy of data analytics, and the value of each to a well-rounded strategic approach:

Data access: This is the foundation on which marketers build by collecting all pertinent information about customers, including name, address, demographic data, history of transactions, product and service purchases, and responses to past campaigns. Every business should earmark the appropriate resources to ensure this data is as accurate and up-to-date as possible.

Reporting/profiling: Key performance indicators are developed and applied to track the performance of customer relationship management (CRM) initiatives over time and across customer segments. Here, marketers can also track client migrations across various segments, compare responders versus non-responders, and gauge campaign response over time.

Current value: The underlying premise for CRM is that not all customers provide equal value to an organization. Therefore, the first step for any CRM initiative is to measure customers by their value to the organization.

For example, 20% of clients might account for 80% of a company’s business, and would be worth a lot of the marketer’s time and money. Another 30% might be designated as moderately valuable, but having the potential to move up into the top 20%; they’d require a different kind of pitch.

The last 50% could account for just 5% of the company’s business; they are less committed, motivated largely by price, and require still another approach (or, maybe, none at all).

Segmentation: In this stage, marketers identify prospects who share similar characteristics – who, therefore, belong to one of several specific segments.

This provides the opportunity to focus on the highest-value segments and acquire new customers who match the segments identified as most desirable. As well, sales pitches can be custom-tailored to suit each segment using what is known about those segments. Customers can be segmented using many criteria.

But segments should focus on identifying customers with similar product and service needs as implied through neighborhood socio-demographic characteristics, life stage, usage behavior, or needs and attitudes as identified by market research.

Predictive analytics: Use this to predict each customer’s likelihood to initiate a particular activity in future based on their unique characteristics and past behavior.

The benefits represent a “win-win” for the organization and its customers, with marketing ROI rising, and customers receiving more relevant offers – the principle of “right message to the right customer.” Predictive models are developed to assist marketing at all stages of the customer lifecycle, including acquisition, cross-sell and up-sell, retention, and re-activation.

Potential value: This is assessed by combining each customer’s current value with their potential to buy more in the future. As with current value, potential value creates an even clearer way to identify the most valuable customers, the ones worth keeping.

It also helps to identify those less valuable customers with potential for entering the most-valuable category, and those low-value clients on whom it may not be necessary to spend as much.

Database marketing strategic framework: Using these six stages, marketers can develop a database-marketing strategic framework that differentiates customers based on the value they currently contribute to an organization, their product and service needs, and their potential future value.

Each segment so identified should be assigned its own distinct marketing objectives. And it is crucial that these distinct objectives are understood across all channels of an organization, including marketing, sales, customer service, and operations, so that all can work in support of them.

Additionally, a strategic approach to database marketing can facilitate consistent communications to the customer across multiple channels, including Print, Web, e-mail, POS and others.

During a campaign, and at the end, it is crucial to have key performance indicators to measure and track results in a segment-specific way. What are the criteria for success?

What will the organization regard as a minimum ROI? These indicators will be invaluable in planning subsequent campaigns, but they have to be developed in advance.

Finally, it is essential to devise ways to capture all of the data, both raw and processed, which went into a campaign and emerged from it. As with the performance indicators, this data is essential for crafting subsequent campaigns; rather than starting from scratch with each campaign.

Rick Brough is director of consulting services for Toronto-based Transcontinental Database Marketing.