Clean Databases Transform Your Business

Today’s CEO needs to start steering toward the transformation from a products- and services-centric business to a customer-centric business. Where does that transformation begin?

The first step is to create an internal customer database. As companies begin to look at collecting a single, inhouse repository of up-to-date, actionable data, why should they take on this challenge?

Collecting, storing, updating and acting on an inhouse, customer-centric data repository is a key factor in:

• Understanding whether customers are leaving or joining, when and how often.

• Beginning to optimize products and promotions to what customers actually want.

• Enabling dialogue with customers based on their actions.

• Rewarding the best customers and rationalizing the worst customers.

• Sharing information with partners while protecting privacy.

• Enabling a true one-to-one marketing and merchandising culture.

• Receiving actionable, viable turnaround time on customer-centric activities.

• Leapfrogging competitors and stealing their best customers.

Once the decision has been made to collect, store, update and act on an inhouse customer repository, there are numerous challenges. The most critical is creating a customer database with an eye toward clean data. Just ask any colleague working with dirty customer data to explain how little he knows about what is really occurring in his customer base, and he will tell you that dirty data are perhaps worse than no data. Having seen attempts to be successful in this endeavor, here are suggestions to ensure your customer database will be clean:

Expect multiple channels. Even if your business today is not delivering across multiple touch points, expect this in the future. The channels of tomorrow include the Internet, cell phones, kiosks and personal digital assistants. Every existing and new channel needs to have the people, process and technology enablers built in for customers to register, receive promotions and purchase. As the customer repository is created, plan for multiple channel entry points for data. Customers will want to flip from channel to channel as they input and update their profiles, as well as dialogue and transact with your business.

Enforce standard profiles. Keep a standard customer profile collection requirement across all collection channels to maintain consistency. The standard profile is a set of required and nonrequired fields that every customer is asked to answer, regardless of channel. This standard must be published within the organization and enforced.

We have seen companies deliver promotions through new channels or new partners, and suddenly the questions asked of new prospects and customers are in an entirely different format. Suddenly, when the marketers access the information to create promotion segments, or product upsells are created, the existing processes and technologies cannot leverage the new names. This problem is easily solved with an ounce of prevention.

Build a single customer database. Perhaps the biggest challenge is trying to maintain and link multiple customer databases. Over time, every company will need a series of e-customer relationship management and customer-facing applications. Though each system may have its own customer database, none of these will map exactly to the needs of the enterprise, or have knowledge of the other systems that need that data.

For this reason, the only right architecture is to create a single, cross-channel, clean enterprise repository of customer information. This hub is called an operational data store. Create a hub-and-spoke technical architecture. At the hub, data must be cleansed, standardized, validated and de-duplicated once. Require the spoke applications to pull from and publish to this ODS in a managed fashion.

Deduplicate strategically. The decision on how stringently to deduplicate is a decision on a spectrum, with pros and cons for each choice. Deduplication is a process whereby the duplicate entries for a single real customer are edited. The result is an accurate representation of the customer universe. Often companies will decide to use a loose method that optimizes for ensuring that they are not losing any customers or prospects. Since this includes fewer editions, the downside is that there may be customers receiving duplicate promotions. On the other hand, using a stringent approach will mean fewer double mailings, but that some customers are edited out of existence. There is no right choice along this spectrum. Consider the choices and weigh the outcome based on business requirements.

Validate. In the past, some channel partners executed address validation. For example, direct mail partners could not be sure of the quality of customer lists received from clients. It was their responsibility to manage data cleansing. As the number of output channels increases, there will be more pressure to push the cleansing process closer to the data source. As a result, consider doing address validation inhouse in order to save costs and help with deduplication.

Know your privacy policy. Privacy statements deserve research and careful consideration. Most privacy statements will touch on how and when the information can be used, at detail and aggregate levels. The first step to success for the data architect is to know your privacy policy. Understand what data can be added within the policy and what data cannot. Know how the data can be used, and when the data cannot be used. Use standard privacy language when possible. Keep an eye toward customer satisfaction, federal and international law and sharing data with partners.

Add data incrementally. The customer repository can and must be continually updated with additional information. Usually, corporations have a wealth of internal silos of data that can be joined to the customer repository. Externally, demographic and psychographic data providers offer myriad possibilities. Partners are yet another source of rich information.

Use the proper key. Although this seems simple enough, it is important. Maintain unique, meaningless customer identifiers for each customer. Often, the customer repository will be tied to an application. The customer identifier should be unrelated to any application, program, customer name or information, collection channel, loyalty card number, database or computer system. This single suggestion is guaranteed to save at least one headache per week.

Commit to customers. Often in the past, customer satisfaction and the customer asset were hyped. Corporations gave lip service but little investment. You will begin to see the customer commitment higher and higher in the organizational chart. You can appoint a chief customer officer. At a strategic level, this role would include building and extending the customer asset. The officer would be responsible for increases in registered customers, net migrations of favored segments and clean customer data. At a tactical level, this role would include the definition of a standard profile, incremental data additions and adding eCRM applications to the hub-and-spoke architecture.

Use it. Perhaps the most important manner to understand how to improve the cleanliness of the system is to get a lot of users and applications accessing the customer ODS. The more users and business and technical requirements that are thrown at the data, the more often its weak points will emerge.

• Paul Springmann is senior consultant at Vision Chain, Reston, VA, which provides eCRM and decision-support software and consulting services to large retail and manufacturing companies. Reach him at [email protected]

Related Posts