Data quality is to the information society what quality of materials and products are to the production society. The quality of data in a business is a deciding competitive parameter, and this recognition is currently spreading among businesses throughout the world.
The technological development has generally, along with the development of ERP systems such as SAP, made it possible for businesses to produce, modify and disseminate enormous amounts of data. These data often form the basis for decision making, financial reporting and controlling liquidity, products, clients and suppliers. Senior managers, investors and stock-exchange analysts have therefore begun to see the quality of the content on the ERP systems as particularly relevant in relation to credit ratings and price risk assessments. Therefore, it stands to reason that data ought to an even higher degree be seen and treated as a strategic asset to the business.
It is our assertion that the quality of data in a business’ ERP system at times can say more about the business’ ability to create profit over time than an analysis of the stock market value. This is supported by research conducted by PA Consulting Group in 18 businesses that jointly represent 17% of turnover in the Danish top 100 businesses. Similar but more extensive research conducted by DM Review based on American businesses indicates that the cost of poor data quality amounts to approximately 20% of business turnover.
The reason for this is that poor data quality causes a great amount of non-productive activities as well as increasing the risk of the management making important decisions based on an insufficient or even incorrect foundation. It simply makes BPM initiatives more difficult to manage.
The costs of poor data quality are extensive but often hide in other activities. Examples such as wrong decisions, faulty client or supplier data, maintenance and storage of data which is never used, early or late payments, internal work time used to dun others for data and information are only quantified with difficulty, which makes the area and the related problems invisible to the management of the business.
This article describes why data should be treated as a strategic asset and makes suggestions to how the business can begin working with optimization and ensuring proper data quality.
BPM and effective data control requires the focus of the top management
Data is generated, modified and disseminated in practically every activity and process throughout all of the business’ value chains. This often has the result that the task of ensuring proper data quality quickly becomes either extremely extensive and resource consuming or too general and unfocused. Therefore, working with data quality management (DQM) in business processes requires the focus of the management as well as a structured approach.
Several organizations have already embraced the consequences of this and have established DQM as a separate area for top-level management. In USA it is not uncommon to see a senior management position with the title of Chief Data Officer (CDO). The CDO’s job is to maintain a comprehensible grasp of the data quality in the financial and administrative IT systems of the business as well as being responsible for quality processes within data and information across the organization. In 2005, Osama Fayyad was appointed CDO of Yahoo Inc. – the first of his kind on the market. According to Fayyad himself, “the goal was to build a bridge between data and business and to explain that data is a strategic asset, not an IT service.”1
A classic example of the importance of this area can be found in the Accounts Payable (AP) process of the business. If there are inconsistencies between the registrations in AP and what has actually been agreed upon in contracts and service agreements, these might mean that invoices are paid too early or too late which again might have a negative effect on the working capital and brand value of the business.
An efficient method for improving data quality in a business is introducing a system for controlling data quality. A data quality control system is a system for regular and structured work on maintenance, improvements and surveillance of data quality across the business. An example of how such a system might be constructed is illustrated in the four steps below:
Step one – identifying business-critical data
In order to ensure data quality in a business it is usually necessary to divide data into more or less important areas. Certain data is not particularly important for the daily operations of the business while other data is outright critical for the business. Here it might be beneficial to work with several individual assessment criteria as shown below:
This work is really a risk assessment of data, which attempts to answer the question “how high is the risk to the company if these data are corrupt or incomplete?”
The first step results in a quantity of critical data areas which should then be an obvious part of the company’s data quality system. In addition, it is possible to further dissect these critical data areas into smaller segments such as transaction data and master data.
Step two – Establishing quality parameters
In our work with ensuring data quality we have often heard leaders complain about the quality of for instance client master data. When we then ask them what the level is, or what an acceptable level would be, we rarely receive a clear answer. In order to work in a systematic and structured manner with improving a company’s data quality it is necessary that the management reach a decision on what good data quality is. For example, can the company live with 5% of the order forms not being correctly filled in? Or should this limit be 10% or 2%? When the issue is safety, regarding for instance handling of poisonous materials, the company typically needs to know the location and condition with 100% accuracy without exceptions.
The result of this work is several parameters that define the level of good, medium and poor data quality for selected business-critical data. These levels can be broken down further, if needed.
Step three – Data control and management
This part of DQM is about authority, procedures for reporting and escalation and delegating responsibilities and duties in the business as related to data quality.
This system is meant to ensure that the company can rely on work being done regarding securing, measuring, reporting and perhaps improving data quality in areas such as critical clients, financial ratios, production data, logistics data, etc.
This might result in creating a CDO position as well as dismantling – or centralizing – the company’s data ownership. The task here is to ensure that there is a clear understanding of who is responsible for the selected business-critical data. Often, the task will include breaking down processes in detailed data flow charts.
The challenge for companies that grow through mergers or acquisitions is often that several different ERP systems are used simultaneously. It might well be very costly to integrate and ensure the use of a joint IT system and it might also be necessary to be able to work with quality assurance across systems. This makes it crucial for the company to create joint quality assurance processes and quality management system across the systems.
Often, an IT integration project will be very costly with a time frame rarely less than six months. A well-equipped DQM project will be able to deliver visible results within approximately three months, and the costs will typically be five to ten times lower than the realized profit.
This step is meant to result in a well-proportioned and structured data quality organization across the business, where duties and responsibilities are well defined according to the selected business-critical data. At this point, the company must be in control of reporting, data ownership, escalation procedures and data quality criteria.
Step four – data quality measuring and audits
During this phase, the selected business-critical data is measured which often results in quite a few mini-analyses within each data area. For this task it might be necessary to employ both qualitative and quantitative methods and at times it will not be possible to measure the area directly, in which case indicative data will be measured instead.
The most essential thing, however, is that when the first data quality measurements for a business-critical area are carried out, a forward-pointing method for measuring quality within that same area is identified and established. The practically oriented measurement must be described in order to become a natural part of the data quality management system.
Working on introducing data quality audits might help to ensure momentum, steady improvement and management focus. A data quality audit is somewhat similar to a financial audit but focuses instead on the consistency and quality of data and normalizing data generating and data modifying processes. Additional focus is placed on whether the data quality system is robust, and whether measurements, procedures, preventive measures and continuous adaptations are carried out. Such a review provides management and investors with a rapport, where data quality is treated separately and which describes the control and management of the business-critical data of the company.
Perspective
Only a few businesses and organizations yet work with improving and monitoring data quality in a structured manner, although we do observe a tendency towards greater use of such systems as more businesses begin to view their data – and not just their ERP system – as a strategic asset.
Not to view data as a strategic asset might be the single most important mistake of any business and it makes it difficult to capitalize on BPM initiatives. IT systems and ERP investments alone do not create the results of a business – the information and the data within those systems equally give a company an opportunity for building and maintaining client relationships, financial reporting and finally the possibility of optimum resource management in the company.
Therefore, most organizations should apply resources towards securing and managing their data as an important strategic asset together with their BPM projects. Effective Data Quality Management can help businesses and organizations become more profitable and successful and needs to be related to BPM initiatives.
1 Gregory Piatetsky-Shapiro: “Interview with Usama Fayyad, Yahoo Chief Data Officer”, http://www.sigkdd.org/explorations/issues/7-2-2005-12/fayyad.html
