The ROI of Data Quality
Enterprise-class data quality solutions incorporate a scalable, repeatable process for data quality management.
Business and technical managers are having a hard time presenting a viable cost justification for implementing data quality in their organizations. This is in spite of a variety of industry studies and polls that attest to the importance of data quality for successful data warehousing and other business initiatives. The problem lies in a disconnect between the individuals that actually see data quality problems and decision makers that need to understand the bottom-line impact of data quality in their companies. This article bridges the gap between those data users and managers, presenting six specific business cases for implementing an enterprisewide data quality solution. From accounts/receivable to call centers to marketing and sales departments and beyond, these cases encompass both bottom-line arguments and real-world examples for data quality. They include examples of the risks of bad data as well as the benefits of a well designed, enterprisewide data quality solution.
Do you have the numbers on data quality? This article discusses the touchpoints in an organization that benefit from better attention to data quality and the risks to profitability associated with lack of attention.
The Bird’s-Eye View of Enterprisewide Data Quality
Your business will lose revenue and reduce operating efficiencies if it doesn’t pay attention to data quality. This is the consensus among more than 600 CIOs and IT directors polled about data management (PricewaterhouseCoopers, 2001). According to the same survey, IT executives have become more experienced at implementing CRM and data warehousing systems. They have realized—often through failure—that high enterprisewide data quality (EWDQ) is one of the most important criteria for project success.
EWDQ unifies disparate pieces of information about any entity in the business sphere to enable a complete customer view. From customers to supply chains to products to physical warehouses, and beyond, each far-flung business entity contributes to the bottom line. EWDQ lets businesses realize and act upon the most accurate and complete information about all business entities, despite disparate systems, varying data formats and international data sources.
Functionally, EWDQ is more than name and address verification or record de-duplication. It is a complete system for defining and enforcing global business rules for customer- and business-data quality. EWDQ processes are the foundation of the business’ ability to reconcile global data with existing data processes and infrastructures, ensuring complete customer knowledge and data transparency throughout the enterprise.
The inclusion of global (e.g., multinational or non-English) data in EWDQ is critical, since international businesses are especially sensitive to the data quality component of enterprise data management. Driven by market forces and the quest for additional revenue, businesses selling to international markets can incur enterprise data overload. As international sales, accounting, call centers, and support teams contribute to the data warehouse, new information content and formats often outstrip existing data management capabilities. The result is massive amounts of unchecked—even unusable—stored data.
EWDQ averts data overload by empowering companies to identify, standardize, verify and enhance data from businesswide sources and to link records based on data elements. The result is an accurate, complete, and relevant view of business entities that is consistently available across all business and customer touchpoints. If you accept the notion that data has value, it follows that EWDQ builds business value in two ways. It allows all business units to optimally use diverse information collected through disparate channels, and it provides each customer touchpoint a unified, accurate customer view. As a result, businesses can:
Given that many researchers believe that the amount of data entering an organization will expand a hundredfold over the next five years, Fortune 1000 companies must increasingly depend on EWDQ processes to maintain high velocity, active customer operational environments with cost-effective data quality management.
Finding the Numbers for EWDQ
Even if arguments for the value of EWDQ seems logical, many companies have a hard time justifying the cost of an EWDQ solution. According to industry analyst Gartner, Inc., about half of enterprises with a CRM strategy are unaware of data quality problems. The remainder, however, recognizes data quality problems but do not see the value of fixing them (Gartner, 2001).
This article answers the value question. Through concrete examples for a cross-section of business units, it provides real numbers for benefits forged by successful EWDQ. Specifically, it examines inefficiencies that can creep into accounting, support, fraud detection, compliance, marketing, and call center efforts. It also discusses data quality’s impact on customer satisfaction, the creative process and “value-added” work, processing revenue, and the actual financial performance of the company.
The first step in assessing EWDQ potential is evaluating how enterprise data impacts your bottom line. Find areas where data use can be given dollar values. This could be the value of direct mail pieces lost due to misaddressing, the value of debts unrecovered because of misdirected bills, the cost of additional time spent on customer support calls due to incomplete customer records, and so on.
This article doesn’t propose to explore every use and impact of data; instead, it presents a few representative cases from diverse departments across the enterprise. In each case, a simple ROI formula provides a template with which you can calculate your own EWDQ ROI. With your real-world numbers, these equations will help justify a unified customer view within your own organization.
The ROI of Data Quality
The ROI of Data Quality
Accounts/Receivables ROI Calculation
The ROI of Data Quality
Case 1. More Accurate Accounts/Receivables
Attention to data quality in your billing system profoundly effects the collection and reporting of revenue. The more incorrect data and other billing errors in your database, the longer it takes to collect revenue. It is easy for an accounting department to be bogged down in billing disputes that erode customer satisfaction. EWDQ helps avert this situation and protects the bottom line from inaccurate billing that might mask the real financial picture. Moreover, EWDQ prevents unnecessary referrals of accounts to collection agencies because of incorrect data.
Imagine a fictitious company, Widget, Inc., in which the accounts receivable department attributes $250,000 in debt one month to non-payment. How much of the missing receivables is due to slow payment remittance and how much is due to data quality?
Widget knows through experience that about 20 percent of records in a given database reflect changes in address data each year. Customer moves, mergers, and acquisitions can wreak havoc on customer data: some business customers might be acquired, for example, and request billing through another party. With this information, it is easy to compare the capital impact of data quality on Widget’s accounting department.
With no data quality system in place, Widget would send its $250,000 in bad debt to a collection agency. This might yield about 25 percent of the debt, or $62,500.
With an EWDQ implementation, Widget could consistently track billing redirects and address changes across all customer accounts, allowing it to collect $100,000 of the debt before sending the balance to its collection agency. Yielding the same 25 percent, the agency would return an additional $37,500.
Comparing total debt recovery between the two scenarios, the bottom-line impact is clear. The EWDQ implementation would allow Widget to collect more than double the amount it could recover through the collection agency alone (see Figure 1).
There are other liabilities to consider when estimating the impact of EWDQ on billing. For example:
Inconsistencies, omissions, and inaccuracies in customer information destroy its reliability and undermine existing systems, technologies and processes. EWDQ relieves these systemic stresses, enabling more efficient processes, reducing time and energy dedicated to poor data quality effects such as unnecessary billing disputes and supporting greater customer satisfaction.
The ROI of Data Quality
Call Center ROI Calculation
Case 2. Better Customer Service
A good customer experience for both the company and customer is one that takes place smoothly, accurately, and—most importantly—quickly. In a recent article, Siebel estimated that the cost to handle a single customer service call can range from $300 to $350, based on the length of the call, the company’s infrastructure, training costs, peak times, downtime, personnel, and facilities needed to handle the call. Better EWDQ lets you achieve more revenue per representative with:
Every second you can shave off call center interactions, saves you money. Over the course of a single year, even a two-second time saving per phone call could add hundreds of thousands of dollars to the bottom line.
EWDQ provides the time savings that translate to cost savings. Moreover, it ensures the complete customer view that opens the door to broader sales opportunities and higher customer satisfaction. As companies turn over to call centers more ownership of self-service processes such as online interactions, the dollar value of interacting efficiently and effectively will also grow.
Case 3. Deeper Brand Penetration
The goal of your investment in developing a respected recognized brand is to drive customers to your points of sale (POS). It would be a shame if you failed to recognize a loyal, profitable customer at a given point of sale? Relationships that have lasted for years can be lost in a matter of seconds when call centers, Web applications and POS representatives fail to identify and understand the relationship with an existing customer or fail to respond appropriately to the total value of the customer across all accounts.
Consider just a small slice of customer service, marketing to existing customers. Clean data and proper analysis can let you increase your yield tenfold. Here’s another example from Widget, Inc.
The company is considering a direct mailing to its entire customer base to announce a new product offering. Since Widget doesn’t have EWDQ, its database is full of duplicate records. It also contains some duplicate mailings to different members of the same household. Skimming the records, a marketing executive can see that Frank Johnson, his wife, Frieda, and his son, Frank Jr., are all listed as separate customers at their address on Main Street. Because of these duplications, the planned marketing program is likely to be a failure, relative to its cost. The additional expense of sending redundant mail pieces to individuals and households will approach and perhaps even overtake any profit potential.
An EWDQ solution would allow Widget to de-duplicate individual customer records. Moreover, it would link the records within the database into households. This would allow Widget to send just one mail piece to the Johnsons and other households in their database. Widget would still get its message across, cut its costs per response and increase its total profit on marketing campaigns.
The result is a cost/benefit analysis for direct mailing that makes sense. Lower mailing costs with a similar yield allows Widget to profit from the marketing program that would otherwise represent a loss.
Better attention to EWDQ also makes direct mail a winning scenario for your customer. When you inundate customers with fewer duplicate offers, your customers feel that you’re paying attention to them as individuals. Better address accuracy also helps ensure that targeted messaging reaches the right person at the right time.
Yet another benefit of the complete customer views afforded by EWDQ is support for more accurate business and customer intelligence. Again, consider Widget and the Johnsons. Frank Johnson orders a super widget, which Widget records as revenue. Frieda doesn’t like the shape of the widget, however, and returns it to the company, which Widget records as an expense. On the whole, Widget loses money on the sale because of processing overhead.
Still, Widget sees Frank as a very good customer, because his record doesn’t reflect the returned item and nothing in the database ties him to Frieda. Widget would continue marketing to Frank. Based on its erroneous understanding of his transaction history, it might even offer him discounts and promotions based on his perceived profitability. All of it would add up to more lost money for Widget, less profitable marketing programs as a whole and less effective branding efforts.
With EWDQ, Widget could reconcile Frank and his household’s interactions, building a complete understanding of the Johnsons’ relationship to the business. Only with such an understanding could Widget respond appropriately to the Johnsons’ total business value.
Marketing ROI Calculation
Case 4. More Unity in Complex Organizations
Market leaders like Microsoft and 3M point to huge annual operational cost savings due to the efficiency of their data quality solution. Cost reduction is a major driver in the efforts of business to better understand customers. For example, because large companies can be so highly diversified and operate in so many locations, they can have multiple sources of trading partner data. In these cases, most large organizations have significant data redundancy, inaccuracy, and inconsistency because different business units serving the same customer often have diverse ways of spelling or representing customer data.
In large organizations, resolving these issues to create high data quality is critical to meeting customer service expectations. The data quality process starts with creating accurate customer names and addresses. This can be a complex process, but it is critical. In some regions of the world, for example, customers might reject entire shipments at the point of entry, if the shipment’s invoice is incorrect. The result is extra work, lost time, additional shipping expenses, and lower productivity—all expensive side effects of poor data quality.
EWDQ and a data warehouse provide opportunities to make data more universally accessible and to cleanse data for uniform representation across business units. The data warehouse allows cleansed data to be distributed to legacy and other operational and customer-facing systems with mission-critical functionality.
EWDQ can be extremely beneficial in multi-divisional corporations. For example, corporations with no way to monitor customer activity across touchpoints may face the unfortunate scenario of sales divisions bidding against each other for business, lowering profits, and threatening customer confidence. EWDQ ensures the organization can understand how it is interacting with a customer, no matter how complex, distributed, and multiform the interactions.
Another impact of EWDQ is on the sales force. A sales force is likely to lose confidence in a prospect database with more than 10 or 20 percent invalid addresses. Sales representatives that use unreliable databases become less productive as they chase more poor leads. As a result, they might seek out database alternatives that further fracture and distort customer views.
In the physical warehouse, poor knowledge of customer buying patterns can lead to inappropriate inventory levels that make costs soar and contribute to earnings losses. It is difficult to track product sales if part identification isn’t standardized. What appears as a KMX –123, KMX123, and K123MX in the French sales database, may appear as WIDGET-Pro and Widg. Pro in the U.S. database. If the corporation tracks inventory using only one or two of the identifiers, it might severely mis-estimate real inventory. Without EWDQ, it might be easier, more cost efficient, and more reliable to ask a given customer what he or she is buying, instead of piecing together disparate information from a fragmented database system with poor quality data—although both are expensive and unattractive options.
Case 5. More Accurate Fraud Detection
The banking industry knows the importance of fraud detection in controlling operating costs. Matching your customer and prospect list against a list of “bad guys” and their aliases is a key step in protecting your organization against fraud.
Many fraudulent schemes involve criminals maintaining several accounts with a single organization. Criminals create and maintain multiple accounts by using multiple name aliases or other identity variations. With EWDQ, you can more easily spot and react to relationships between records and accounts by the same individual in multiple divisions that might indicate fraudulent activity. Here’s an example from Widget, Inc. Bob Smith registers for an account using the following information:
Within a year, Bob defaults on $10,000 in payments. Soon thereafter attempts to open a new account as:
Robert K. Smith
In many databases, even the minor variations in name and address information would allow Bob to again become Widget’s customer. Widget might say their prices are practically a steal, but in Bob’s case, the lack of EWDQ ensures that they really are.
Federal agencies such as the Treasury Department issue a list of suspected criminals (such as the Specially Designated Nationals and Blocked Persons list from the Office of Foreign Assets Control, (http://www.treas.gov/ofac/) that companies can and might be required to match against their customer databases. Only a robust data quality tool can match complex name forms that overwhelm simple searches (see Figure 3). Moreover, only an organization with EWDQ can ensure that the matching process includes all customer records.
There is no limit to the vulnerability of your organization when it comes to fraud. Consider the effect that an unsavory character may have on your insurance rates, credibility, and liability. EWDQ is an effective weapon against fraud, increasing your ability to identify criminals and criminal behavior.
Case 6. Increase Strategic Impact
Productivity. Business intelligence (BI) systems help instill knowledge of your business and sales processes. Raw customer and business data can be unreliable, however, and additional checks and balances are necessary to insure accurate analysis. EWDQ lets you rely more on BI reports without chasing down missing information or reviewing paper-based records.
But how accurate is accurate enough? If your company isn’t thinking about data quality, it could be undermining profitable opportunities and repelling loyal customers—even if data is 90 percent accurate.
For example, given the inaccuracies in Figure 4 above, how many widgets did U.B.M. buy last year? Was U.B.M. treated as a top customer last year or just one of the crowd? If your reports miss just one of U.B.M.’s transactions, it could be missing as much as 20 percent of business in projecting sales from the company this year. Loss of confidence in the system devalues it, leading to a manual check of the database, a cross check against billing records and other productivity leaks.
Going Concern. Strategically, businesses must face the onslaught of data flow while thinking about the “going concern” of the corporation. Most businesses experience IT turnover. Experts agree that enterprises will experience turnover rates of at least 20 percent annually among key knowledge and leadership workers. A well documented and enterprisewide system of business rules helps the company to withstand such intellectual turnover with minimal costs. Data quality solutions are the best way to implement and maintain a process that will continue to provide with data that withstands the evolution of the employee bases, the corporate direction, and even the market.
Data Quality Solutions: Three Approaches
Data Quality Deployment Costs
Some organizations take the manual route, hiring staff to manually scour their databases and de-duplicate records. Such manual efforts, however, are relatively expensive and don’t provide record linking. Training costs and challenges involved in making sure staff learn and follow business rules are only the beginning of the investment. As databases grow and diversify, manual solutions become exponentially less efficient and effective. Finally, most such homegrown solutions aren’t easily managed across multiple data systems.
A more common approach is launching an in-house programming effort that specifically targets record linking and postal data correction. Such solutions often provide good-news-bad-news results. On one hand, the database programmer will never be out of a job; on the other hand, the payments for that job will never end! The diversity and complexity of data quality issues means that a homegrown solution might eat up programming resources for months or years. Homegrown solutions also generally lack the data profiling functionality of EWDQ software solutions. Without clear insight into data characteristics, it is nearly impossible to effectively address the unique challenges of diverse corporate data.
In the final analysis, a software solution that analyzes, standardizes, cleanses, enhances, and links global customer information is often the most economical and effective solution. An enterprise-class software solution leverages vendor expertise in data management to augment and empower the capability of data quality staff. EWDQ software provides the transparency that helps preserve an organization’s going concern.
Enterprise-class data quality solutions incorporate a scalable, repeatable process for data quality management. This key feature is a critical distinction over most homegrown and manual solutions, taking the business beyond localized data cleansing. Scalability and repeatability enable rapid initial implementation, phased deployment against concrete objectives, and consistent process implementation across the entire organization.
Picture your business two, five, and 10 years in the future. Now, focus on your prospect, customer, and transaction databases. Will they be full of valuable, timely information or a jumble of duplicate records and misleading data? Most experts agree that businesses will increasingly rely on worldwide data to make decisions and market products. Poor data quality does not stifle this trend, but it does impede competitive position as you place more value in organizational data. The ROI examples presented in this article demonstrate that an EWDQ solution can help you gain and strengthen both strategic advantage and your bottom line.
PricewaterhouseCoopers. Global Data Management Survey, 2001, PricewaterhouseCoopers, Boston, Massachusetts, 2001. Gartner. “Customer Data Quality and Integration: The Foundation of Successful CRM,” Gartner, Inc. 26 November 2001. Laney, Doug. “Customer Data Quality: Part 1 – A Taxonomy of Troubles,” META Group Research, Stamford, Connecticut, 19 June 2000.
Len Dubois - Len Dubois is the head of marketing for Trillium Software. Trillium is the leading provider of data cleasning and reengineering solutions for on-line and legacy system projects. This is his first authored article. The Trillium Software System has been architected to meet the needs of companies looking to implement enterprise data quality management solutions. You can obtain more information about the Trillium Software System at their website at: www.trilliumsoftware.com.