Customer data is a key organisational asset - perhaps the most valuable asset an organisation owns. Board rooms recognise the value of customer intelligence and sanction investments of millions of pounds in data warehouses, customer relationship management (CRM) and other customer-facing IT initiatives. They do so to support improved governance, more profitable and competent customer interaction, better marketing and other key processes.
Risk management is often associated with the physical aspects of doing business. But just as system failure risk is a consideration of risk managers, so perhaps should be enterprise data quality risk. If managers, and the processes and businesses they govern, are to operate effectively and without disruption, then we must ensure the data that supports them is of sufficient quality. The information must be known to be 'fit for purpose', in terms of its integrity, accuracy, completeness, timeliness, uniqueness and relevance.
The fact that data warehousing has grown into a £3bn global market is demonstrative of the drive organisations have undertaken to pro-actively manage their data and extract information from it. Only the most insightful boards, however, have recognised that the return on investment (ROI) from such initiatives is subject to the quality of the customer information which is fed into them.
Data quality and risk
Spend several hundred thousand pounds on implementing a CRM system and feed it with garbage, and it will provide your management, your marketing and sales teams and your contact centres with rehashed garbage. The deployment will not be accepted by users. Business will be disrupted and the investment will have failed. It also means the company will have lost valuable time in its race for improved processes and competitive advantage.
Perhaps you are confident that your IT group will certainly have data quality in hand when implementing such projects? Then consider the success rates for CRM deployments: 'Through 2005, more than 50% of CRM deployments will suffer limited acceptance, if not outright failure, because of the lack of attention to data quality issues'. (Gartner)
The results of poor data quality can be disastrous, yet many IT functions do not seem to consider the issue until too late. Despite the disruption that low data quality can cause, management do not appear to recognise the extent of the problem either. According to a survey by the Chartered Management Institute, respondents asked to consider IT risks cited their concerns as being loss of IT capacity, loss of skills, people, negative publicity and fire - no mention of data quality there.
It is not just CRM projects that suffer limited success due to data quality issues. Supply chain management (SCM), enterprise resource planning (ERP) and other projects requiring merging of data sources are also fraught with risk. According to Gartner, 'Most information engineering initiatives will fail due to a lack of data quality'.
It is not only the successful ROI of IT investments which are at risk.
Executives basing decisions on unreliable data will make bad ones, despite using their best judgment. This leaves the future of the business somewhat to chance, and can also expose directors to charges of poor corporate governance.
Financial institutions need to consider whether customer data used for assessing credit risk is suitably accurate too. They must also judge whether efforts to provide accurate risk information in accordance with The Basel II Accord could be exposed to data quality risks that might undermine compliance and timely delivery to budget.
Poor data quality undermines the success of tactical activity such as marketing campaigns, and annoys customers. In most industries customers need only be badly treated once or twice before they will look elsewhere.
The ROI from merger and acquisition activity can also be affected by data quality. In addition to acquiring physical and intellectual assets, acquiring customer databases is often a strong attraction. Almost inevitably, in most cases, there will be a degree of cross-over between the customers of company A and those of company B. The extent of this cross-over, if it were known with some accuracy, might influence negotiations. Then, after the merger or acquisition, the value of what is hidden in the data must be properly evaluated and rapidly leveraged. If the acquiring firm does not itself have strong data quality, consistency and standards, and has not developed a tested method for comparing and merging customer data from dissimilar sources, then an acquirer may end up with little more than discouraged employees and a surfeit of real estate and capital equipment.
According to Current Analysis, a US-based analyst group, the easiest way to demonstrate that data problems can occur is by illustrating some common problems that affect data and risk the value of CRM investments, governance, compliance and customer relationships. At a simple level, misspellings and typographical errors are common, as are incorrect data types (for example alphabetical characters in numerical fields or the number 20 in a field for the month) and homonyms (words spelled alike, but which have different meanings).
There are also problems regarding standards that arise when different business units enter data either into the same, or different databases.
In such cases, duplicate records can occur when the same individual is variously addressed.
Moreover there are frequently problems with missing, or so-called invisible, data. A record may look correct, but, because there is some piece of information missing, there are problems with links and identifications. For example, the address '2 Century Place' may be correct, but lacks an office location to segregate it from the 50 other offices in the same building. Likewise, the name Terry Jones is correct, but without a gender designation, accurate matching is less certain.
These examples may appear low-level stuff, but the frequency of quality problems often surprises even data analysts, and the complexity of resolving them in multi-million record databases is very rarely appreciated. The impact of such errors can be dramatic.
Cause of the problem
Data quality problems arise for a variety of reasons, but they are easily understood.
- Real-time data entry procedures are not followed and controls are insufficient. Over several years, this can lead to multiple files for the same customer, which can be difficult to track down. Businesses with contact centres and large sales forces expected to enter customer data frequently are particularly exposed.
- A lack of common standards, formats and quality in source systems (often legacy systems) which have later been merged to more modern applications and data warehouses, with imperfect conversion processes. Businesses which have grown through acquisition and those which have consolidated business processes or otherwise integrated large data sources are especially exposed.
- Use of data from external sources which may not always be of sufficient quality, such as regular purchases of large mailing and lifestyle data. Organisations with strong direct marketing functions are most at risk.
- Rapid growth of customer numbers and associated data can mean that suitable processes and controls for data quality are difficult to define, establish and enforce. Companies which have launched successful new businesses and products in recent years are most exposed.
Towards a solution
Most IT directors will be able to describe tactical projects to improve data quality to meet the demands of specific business initiatives. These will typically involve checking the accuracy of customer or supplier contact data, such as names, addresses and telephone numbers. This is fine for one-off tactical projects such as a smaller CRM or procurement initiative, or for cleansing a marketing database. But it is hardly a holistic method of strengthening the data quality platform upon which future initiatives can be built. Tactical approaches do not address the root problem.
What is obviously needed is an enterprise approach, commencing with a data audit of large samples from several sets of key inter-related data sources. A data profiling software solution will identify the frequency and exact location of issues, not just in samples, but in very large production systems, and take a copy for analysis. For example, my own company has undertaken such exercises for BT Retail and Abbey bank. The risk of severe business impact and the urgency of a need for resolution may then be discussed with business decision makers.
Following a data audit, an enterprise data quality solution might then be applied to the sample database sets, to show how much more quickly and efficiently data quality can be addressed by automation, rather than by manual approaches. For example, BT Retail and The Carphone Warehouse use the Trillium software system in tandem, supporting an end-to-end enterprise data quality process.
Rectifying enterprise data quality issues may not be the role of the risk manager. But where data quality is so central to success, the risk manager must surely understand the level of exposure. Risk managers can then present this exposure to the IT directorate and the board. Together they can review the value attached to customer information and whether management has a strategy for enterprise data quality measurement and delivery in place.
To take full advantage of customer data, an organisation must recognise that it is an asset, and that failing to have a data quality strategy will undermine the future strength of the company. Only by first recognising the importance of a solid data foundation, can management finally become open to managing enterprise data quality risk.
In an economy where every customer counts, compromising their loyalty, even unknowingly, is not an acceptable risk. Ensuring data quality across your organisation is critical to gaining an honest and unified view of your customers and helping to secure long-term business success.
Tom Scampion is vice president, EMEA, Trillium Software, Tel: 020 7529 8333, E-mail: tom.Scampion@trilliumsoftware.com
IS YOUR ORGANISATION EXPOSED TO DATA QUALITY RISK?
A few questions targeted at the right people in your organisation will reveal whether they are feeling the pain of data quality issues.
- Ask the IT manager "Is our customer data accurate?"
- Ask the sales/CRM contact centre manager "Do you know the quality of our customer data?"
- Ask the marketing manager what proportion of direct mail is returned as undeliverable.
- Would sales, marketing, customer service, finance and other departments all be able to provide an identical list of your most important customers?
- Ask the sales and marketing managers whether you need to link customer interactions in order to discover cross-selling and up-selling opportunities.
- Ask the IT manager how big an investment the company will make in CRM, ERP, SCM and data warehousing over the next three years.
- Consider how far your organisation is an amalgamation of acquired businesses, with their various formats and standards for data.
- Ask the IT director to explain the corporate strategy for enterprise data quality.
Answers to these question should indicate whether everything is under control, or whether there are risk exposures which might be judged through a data quality audit.
TAKING CONTROL OF CUSTOMER DATA QUALITY RISK
With a data quality solution, companies can create a centralised source of clean and relevant data on which to build a unified, accurate view of their customers. This will reduce the risks posed by unfit data to multi-million pound CRM investments, sales revenues, marketing spends and customer goodwill. Poor quality customer information also risks exposing directors to breaches of corporate compliance legislation, through unknowingly trading with blocked parties or through false declarations.
Here are 10 best practice ideas for a strategic approach to enterprise data quality to review with your IT director and all those owning customer-facing responsibility. They could help make your data work for your business instead of against it.
CONSIDER THE VALUE OF DATA QUALITY. Define the role of customer data within the business. What would the impact of declining data quality be? Have you any quantitative means of monitoring data quality over time?
ALIGN BUSINESS AND IT EXPECTATIONS. Business users are the true beneficiaries of clean data, since most data quality projects help them interact with customers more efficiently and effectively. A cross-functional team of information technology (where the data technologies typically reside) and business users works best to define expectations and data quality's role in the organisation.
CONFIRM SENIOR MANAGEMENT BUY-IN. An organisation that views data quality as a strategic element that continuously enhances and validates information, while preventing corruption of its most vital asset, customer information, will have greater success.
ENSURE THAT BUSINESS GOALS DRIVE FUNCTIONALITY. Align business goals with the functionality needed to ensure success.
UNDERSTAND THE COSTS OF BUILDING A SOLUTION IN HOUSE. It can take tens of thousands of hours to write business rules that address the diversity of words, patterns, phrases, addresses and idioms that comprise business and customer data. Know whether your team has the resources to tackle this in-house, with little risk of failure, or whether a packaged solution with built-in data quality best practices is the solution.
COMMIT TRAINED PERSONNEL. Trained data quality experts, motivated by the responsibility of turning business requirements into lasting data quality standards, are generally more effective and less risky than part-time programmers whose job is to fix each glitch.
UNDERSTAND THE REAL COSTS AND CAUSES OF POOR DATA QUALITY. Experts say data quality issues account for a data warehouse failure of up to 70% and contribute to a 50-70% failure rate for CRM projects. A proactive, enterprise-level approach to instituting best practices in data quality can eliminate risks of business problems that result from data corruption, omission, inconsistencies and other flaws.
EMPLOY A PROVEN METHODOLOGY. By analysing the organisation's over-arching strategy, by paying attention to global language and cultural considerations, and by determining the technical requirements for tracking ROI, each step in the data quality plan produces useful business information for end users.
USE A PHASED ROLLOUT SCHEDULE. An initial implementation within a department or business unit with identifiable, quantifiable data quality issues can yield immediate, tangible results. Implementing a solution with significant out-of-the-box functionality can be another way to score quick wins. Both procedures increase acceptance throughout the organisation, increase support for future rollouts and enable easier risk management.
TRACK ROI. Measure the costs of poor, as well as the benefits of improved, data quality. Shorter processing times, shorter sales cycles, more accurate analytics and higher cross-sell and up-sell volumes all reveal the benefits of improved data quality. Customer surveys can help measure the impact of data quality on customer satisfaction.