Building a Bulletproof Data Mart
The dot-coms that spent the money to develop a solid marketing database are thriving. One thing that has become apparent as a result of the dot-com bust is that the Internet is not a good sales generator. It is, however, an excellent platform for information, customer service and, most importantly, the bottom line. The Internet allows a company to realize major cost reductions and feel the effect immediately.
But before any of these benefits can be realized, it is critical for a company to have a great database under its Web site or to have an e-mail system capable of highly personalized information that can be instantly shared with the customer and all authorized employees who come in contact with the customers.
On the surface, this process sounds easy. In reality, it is anything but. To have a fully integrated database that takes all the various feeds from legacy and Internet systems and seamlessly processes the data on a real-time basis requires discipline and an artful construction of a data mart.
A data mart is quite different from a data warehouse and a customer database that may deal with line-item transaction or behavior information. A data mart most often uses summarized data, meta data or most current customer information. To accomplish this, there is an important need to integrate the data.
Integrating databases: the basics. Integrating data from disparate databases is a complex process. But knowing the elements that make up a successful integration will make it more efficient and effective. The following seven basic steps are needed to integrate data from disparate databases into a unified one.
• Start by being a data detective. Perform a needs assessment to identify all of the databases to be integrated, find out who created each one and gain an understanding of the various platforms and data formats.
• Prepare for the actual process of data integration. Develop a new database that can accept information from various other databases' field formats and will be compatible with the unified customer database you are creating.
• Standardize the data. Because different databases have different conventions for each field, make sure the customer information fields match. For example, change all references to the word "Street" to "St." and ensure that state identifications are standardized.
• Perform a National Change of Address check. The U.S. Postal Service usually keeps data on people who have changed locations for about two years. The NCOA ensures that all the address information is correct and up-to-date. This is primarily done with consumer publications.
• Household, or dedupe the data. Because this step requires more technology and experience, it is usually conducted by outside service bureaus. This process takes various entries in different files (for example, Robert C. McKim, Robert McKim, Bob McKim and R.C. McKim) that may or may not be at the same addressing locations or companies and eliminates duplicate entries. This is a technological process: algorithms comb through the data, recognize similarities and get rid of duplicates.
• Append the addressing information to existing transaction or promotional information. You may also want to append data from third-party sources to provide a more robust picture of a particular record.
• Finally, put the data into a unified customer data mart that has been built specifically to manage the data that has been standardized, householded and appended. This allows the data to be served to the user accurately and routinely.
The investment. The process of integrating data into a unified customer data mart can take three to five weeks, and building the database itself can take up to 12 weeks. However, many steps of the integration and database construction can overlap. For a larger enterprise, which may have the need for internal communication and consensus building, up to three months may be needed to complete a unified customer data mart. For smaller organizations, the process can be completed in about a month.
If you are doing the job inhouse, dedicate at least one person from your staff for every 50,000 records to be integrated. In terms of cost, expect to spend a minimum of $100,000 for software and hardware. Because data integration is an ongoing process, the costs continue after the unified customer data mart is operating. It is important to continue to standardize and household data as it comes in. Overall, the ability to create additional targeted personalized communication and save money elsewhere will offset the costs in the long term.
Improving the bottom line. There are two primary benefits to integrating your data into a unified customer data mart.
First is the immediate reduction of marketing or operational expenses. If the data mart is the backbone of the online systems supporting customer online interaction, the customers will update or correct their own information via an invitation to update their personal history in exchange for a value proposition. The cost for customers to update their own information is 1 to 2 cents per transaction, whereas the cost for a teleservices person is about $6.50 to $9 per customer update.
Customers generally provide a more truthful delivery of information assuming that the questions asked are relevant and not intrusive.
The second and longer-lasting benefit is the personalized communications that result from communication that is timely and relevant. Whether using online or offline interactions, customers appreciate the more personalized attention to their needs.
Though it is a substantial investment, creating an online customer data mart will provide any company with a variety of business opportunities. These range from the potential to decrease operational or marketing expenses to elevating the lifetime value of customers. As such, the online customer data mart is a natural expansion of your data warehouse or customer database.