In practice, though, enterprises tend to focus much more on CRM processes--the "where should this go and who should handle it"--than they do on the actual quality of customer data. "One of the hidden dangers of customer relationship management is enabling better access to bad data," say Erin Kinikin, vice president with the Giga Information Group. The result is akin to putting rotten peaches in faster, better-routed trucks--the peaches are still rotten when they get to market.
Customer data degrades at a rate of about 2 percent per month, which translates to almost a quarter of the entire customer database annually. Typos, misspellings, abbreviation mistakes, and other errors frequently arise because poorly paid and poorly trained data-entry workers in call centers and mail receiving stations have little time to process a large number of orders. Duplicate entries are introduced when the same customers buy products from different parts of an organisation--each with its own system that employs different conventions for organising data. Customer contacts are frequently uncoordinated across channels and divisions, resulting in fragmented customer records and multiple instances of the same customer. Once customer names, addresses, and other key information are misspelled, duplicates are difficult to identify and remove.
Poor data quality causes a number of problems for CRM. Bad customer data can prevent a company from contacting customers when needed, leading to misdirected marketing materials and service notifications. Worse yet, bad customer data can mean that bills don't get paid because they are sent to the wrong customer or the wrong address. Inaccurate order information or customer preferences can lead to extra costs for handling returns and billing corrections--while irritating customers in the process.
Beyond these tactical problems, poor data quality can also undermine the strategic value of an enterprise's CRM system. Without correct data, enterprises don't have a clear picture of who their customers are and how they behave. CRM's big promise is to produce exactly that kind of picture so enterprises can detect patterns and adjust their merchandising efforts to retain valuable customers and increase their spending. "One company had to ask their customers how much business they did with the company, because the revenue was scattered across so many systems and so many subsidiaries with no way to consolidate because of all the different company name spellings," says Kinikin.
Although it's impossible to entirely eliminate data-quality problems, enterprises can certainly take measures to reduce the frequency and magnitude of data errors. Enterprises can begin by trying to prevent data-entry errors from creeping into the system in the first place. Here's a partial list of preventative measures:
- Set up screens so entry clerks choose from a menu of pre-configured options whenever possible--for example, selecting a state from an alphabetised list of abbreviations rather than typing it in. Require approval from the customer or certain authorised departments before clerks can change some types of data, such as a billing address or service preferences.
- Configure the system to check for a duplicate whenever a new customer record is entered. Every time a new record is entered, the system can search for similar records and ask whether the new record matches that of an existing customer.
- Use consistent and meaningful defaults to handle exceptions such as "not applicable" so missing information can be systematically evaluated and corrected.
- Employ a system that properly characterises customer distinctions that matter to your business. A good system permits sales people or call-center agents to record complex relationships and roles rather than forcing them to choose among inaccurate and overly simplistic categories.
Prevention helps, but enterprises still need to have a regular program for correcting bad data that inevitably makes its way past initial entry defenses. All data must be periodically reviewed for errors and duplicates so that corrections can be made. Different software vendors handle different parts of the data-cleansing process. Firstlogic offers software that helps detect data-quality problems. Ascential and SSA match duplicate customer records. Trillium and Vality clean up erroneous names and addresses. Enterprises can also use services from vendors such as Acxiom, Dun & Bradstreet, and Harte-Hanks, which examine data sent to them for duplicates and corrections.
In the end, CRM is only as good as the accuracy of customer information. The problem with inaccurate data is that it can result in a CRM system that does more harm than good. "You can sell new products to new parents as long as you're sure that this 'John Smith' is the one whose wife just had a baby," says Giga's Kinikin. "Otherwise, you just look silly and out of touch."
