Customer relationship management (CRM) was supposed to transform companies as we know them. Business executives drafted CRM strategies and purchased software as they eyed more personal and efficient interactions with customers and prospects. But several years later, progress has been slow, and success has been elusive.
Why? Several reasons explain the failure: resistance to change, lack of software integration and corporate politics, among others. Increasingly, though, the poor quality of customer and other data is cited as the main cause of CRM woes. An August 2001 Gartner, Inc. report titled “Seven Key Reasons Why CRM Fails,” based on Gartner’s research of hundreds of CRM projects, lists the number-one reason CRM flops as “data is ignored.”
Companies including retailers both on the Web and in the store are discovering that CRM is only as good as the data that powers it. Many have skipped ahead in the CRM time line, focusing almost exclusively on installing software for sales force and marketing automation. To get CRM initiatives back on track, Gartner recommends that companies craft a “data quality strategy” and devote “one-half of the total time line of the CRM project to data elements.”
A Three-Step Approach to Improving Customer Data Quality The task of overhauling corporate data can be overwhelming; therefore, it’s necessary to address the process in three general phases.
First, you need to conduct a survey and an assessment of all your databases and the data inside them. This involves an inventory of data types and behavior, an assessment of your expectations for this data and a discovery of specific deficiencies. Next, you must determine which problems are most troubling, and design and execute a plan to fix specific data problems and their causes. Finally, you must perform rigorous, continuous data management through management of both current and new data.
Step 1: Data Survey and Assessment Two major obstacles to good data quality are poorly designed databases and data trapped in database silos. Thus, an inventory of your data must start with a survey of your databases to determine what purpose they serve and whether they are integrated with each other.
In any assessment, the key to success is asking the right questions. For example, is the database you use to process bills the same as your customer database? If not, is it connected to your customer database so when a customer change of address is recorded in your customer database, your billing database is immediately updated? Where does the customer information you collect from your Web site reside, and does this data content feed into your offline repositories?
After you catalogue your databases and map their integration, you must then examine the data content within. Start again with database design. What do the tables within the database represent? Is each table clearly defined? What is their source? How are they updated or refreshed? How comprehensive is each table in scope? What meaning is assigned to each field within the table? Has that meaning changed with new usage?
Then look at the data content itself. How complete or accurate is the data in each field? Are any fields missing, such as work phone number? Can you determine the currency or freshness of the data? Is customer data properly integrated across company databases? Does your company collect the right data it needs to build loyalty through customer knowledge?
You must also determine how your data actually behaves. It may now be significantly different from the intent of the original designer. For example, a “status code” field may contain values different from what is documented in the field’s original set of valid values. This could be attributed to data entry errors, programming errors and spillover from another field, or it can signal that someone is using the data field for a purpose other that its original purpose. Misunderstanding the definition, accuracy or completeness of data can distort the reports created from it and lead to ill- informed management decisions.
The knowledge you capture about databases, tables and fields must be recorded and maintained somewhere in a formalized manner. Large, progressive enterprises put this into a meta data repository, a specialized database containing knowledge about the data asset. This is not a one-time activity. As your business evolves to meet market demands, the scope and complexity of the data required to run your business also increases. The entire data inventory and assessment must be guided by an in-house data steward. A data steward is an executive whose chief role is establishing and protecting the quality of the data asset. This person must understand both the nature of data and your company’s business and determine what data your business needs to perform successfully. Above all, the data steward must treat data as a corporate asset and continually measure its value. If your company does not have someone who fits this bill, make it a priority to appoint one.
Step 2: Data Quality and Process Improvement Once you have identified your most serious lapses in data quality, you need to “mop up the puddles and fix the leaks” in other words, correct the latent data you have and improve the processes which allowed bad data to be captured in the first place.
Incomplete data issues may be resolved by finding new sources for missing fields. Inaccurate or outdated name and address information can be fixed using data cleansing software and services that are widely available. These range from software that can identify and fix common typos and update your records with change-of-address information from the United States Postal Service to sophisticated services that match your data against an up-to- the- second consumer database.
One process that is perhaps the best test of data quality is customer data integration. Try linking all your customer data together an absolute imperative if you’re attempting CRM and you’ll realize how much customer information is scattered throughout your databases and how much duplicate information you maintain.
For example, you may discover that you have three separate records for customer Jane Smith: one as Jane Smith, one abbreviated as J. Smith and one misspelled as Janet Smith. The process of customer data integration will identify these near matches as well as exact matches and serve as foundation for making data-cleansing decisions.
The record for Jane Smith may contain old address information, but new address information may be in the record for Janet Smith. It just happened that when Jane Smith phoned in her new address information, the customer service representative mistakenly created a new record with a misspelled first name. The record for J. Smith might be a Web entry containing product preference information that should added to Jane’s customer record.
The customer data integration process will consolidate these disparate data into a best record for Jane Smith to create a unified, accurate view of your company’s relationship with Jane. A single view of each customer is the fuel that powers CRM software and analytics.
Once your customer data is cleansed, integrated and enhanced with necessary new sources, you must thoroughly document what was done for the benefit of downstream data users. Some data may be less reliable than others, but not totally useless; it should be documented as such so potential users may exercise appropriate caution.
Improving data quality also requires investigating and fixing your data collection processes. This may require programming changes to data-capture screens, valid value tables and other software processes. For example, pull- down menus can be used instead of free-form data entry to guard against typos and data entered into the wrong field. In a Web environment, additional guards against accepting bad data must be built into the application software. Never underestimate the creativity (and potential fraud) of data coming from a Web site.
Step 3: Ongoing Data Management Upgrading the quality of your existing data is daunting enough, but soon updates arrive. Thus, it is vital that you establish a system of surveillance of new data. Software and services that apply data-cleansing procedures at the initial point of data entry are increasingly available to prevent inaccurate data from polluting a company’s database systems.
As a customer service associate, for instance, keys in new data, that information should be verified against internal and even third-party data. Inaccurate data can be automatically updated, and clean data is then introduced into a company’s database systems.
You must also establish data advocacy at your company, turning customer data into one of your company’s most valuable assets. You’d be surprised how few companies do this. The data steward should lead this effort by maintaining an ongoing survey of the data, its meaning and its quality. But many people throughout the company may share a stewardship responsibility for their data. Those who influence the quality of data they capture or maintain, must be properly motivated to constantly improve their processes and watch the quality of their own data.
Achieving customer data quality is the first the step in building a successful CRM program. It is also an effective means of building internal support for a customer-centric business. If every person in your company handles customer data with care, they’ll likely treat customers the same.
1. Ferrara, C. and Nelson, S. “The Holy Grail of Retail.” Gartner, Inc. June 2001.
2. Kirkby, J. and Nelson, S. “Seven Key Reasons Why CRM Fails.” Gartner, Inc. August 2001.
|Sidebar: The ABCs of Data Quality|
Accuracy: The characteristic of the data correctly describing the reality it is understood to describe. This is the fundamental characteristic of data quality.
Behavior: What values are observed and how they change.
Completeness: When a table or file has all the rows or records we expect, or when an individual column has valid and accurate data values in all the cells.
Currency: The recency and reliability of the observation about reality which the data conveys.
Integration: Bringing data from diverse sources together in a way that is consistent and logical.
Richness: Having more data fields. Having more rows or being more complete in covering the scope of reality which is expected.
Register or login for access to this item and much more
All Information Management content is archived after seven days.
Community members receive:
- All recent and archived articles
- Conference offers and updates
- A full menu of enewsletter options
- Web seminars, white papers, ebooks
Already have an account? Log In
Don't have an account? Register for Free Unlimited Access