Blogs

Approaching Data as an Enterprise Asset

Jul 25, 2019 6:35:14 AM / by Abhishek Kamboj

If you walk into a meeting with all your senior executives and pose the question:

“Do you consider and treat your data as an Enterprise Asset?"

The response you will get is:

“Of course we do.”

The problem in most organizations, however, is that while it is recognized that data is a corporate asset, the practices surrounding the data do not support the automatic response of Yes We Do.

What does it really mean, to treat your data as an enterprise asset?  If we were talking about office equipment, corporate offices, fleets of vehicles or many of the other tangible things a corporation may consider an asset, you would hear things like:

  • Planning
  • Acquisition
  • Operation and Maintenance
  • Improvement
  • Monitoring
  • Disposal

When we consider the practices a typical corporation performs for the data assets, we would probably find a similar list of things, but it will most likely read more like:

  • Some Planning
  • Acquisition
  • Operation and Some Maintenance
  • Little Improvement
  • No Monitoring
  • Some Disposal

Is this a harsh assessment of how data gets treated at the enterprise level?  For some organizations, maybe, but they're probably more the exception than the rule.  It's not surprising, though, that this is the reality for many organizations, as the problem is one that has grown over time as the advances in technology and sources of information have grown at rates far greater than our ability or desire to implement change.  In other words:

You Are Not Alone!

The Paradigm Shift

Managing data as an enterprise asset requires a fundamental shift in the way a typical organization operates.

Today many organizations are divided into silos with each responsible for its own line of business or business function.   Communication and cooperation across the silos is sporadic at best.  Decisions while having the enterprise in mind are basically self-serving, with only limited contribution or support for the other silos.  This self-serving nature contributes to data sprawl, adding to complexity, duplication and additional cost for maintaining the data assets.

IT projects are justified, funded, and managed within the silos that need the benefits provided by the projects.  Projects can have cross-silo benefits and responsibilities which are typically evaluated in terms of that silo's benefit and cost.

To manage data at the enterprise level, the enterprise has to change, learning to break down these silo based barriers and create an environment where decisions and actions are in support of the entire enterprise, not the silo.  To achieve this level of cooperation requires executive level backing not only to start the initiative but also as an on-going process to address issues as they arrive in the future.  While making the silo go away seems to be the impossible task, executives can foster the enterprise view by making funding available for enterprise wide initiatives which benefit the enterprise.  Enterprise stake holders can form committees to provide governance and oversight for enterprise initiatives.

How do you get your executives excited about approaching such significant change to support data at the enterprise level?  Relate the opportunities to executive hot spots:

  • Regulatory  - requirements for privacy, data retention, government reporting
  • Cost reduction – for IT projects, productivity improvements, business agility, data sharing
  • Data quality – providing cost reduction, customer satisfaction
  • Intellectual property – supporting discovery, retention and propagation
  • Analytics – supporting improved decision making, fraud detection, increased revenues

Planning and Governance

Data governance is one of the key requirements for successful data asset management at the enterprise level.  At its simplest, data governance is about the processes that control the creation, accessing, sharing, using, and retiring of information, and what happens when there is a problem.

Data governance should:

  • Develop a strategy.  Decide what data to manage,  identify critical and master data.  Determine the value of data based on cost of collection, maintenance, business value, risk of lost or inaccuracies.
  • Establish a committee of the various corporate entities or lines of business that are in a position to understand the data, speak for their line of business, and be able to make a decision
  • Establish a set of policies to define data integrity, quality, security, classifications and use of data
  • Establish a set of standards to control how to implement the policies, like naming conventions, data modelling, tools, technologies, and methodologies
  • Procedures for addressing quality issues, business rule issues, data naming issues, and security issues
  • Exception management and remediation
  • Integration of governance into IT project management cycles with check points to provide continuous oversight
  • Establish a set of penalties for noncompliance (enforced governance requires both rules and penalties)
  • Documentation and metadata requirements

Shared Data Access

To successfully manage data at the enterprise level you need to manage the creation, maintenance, and inquiry of that information from shared applications.  Many organizations are looking to Service Oriented Architecture (SOA) as an approach to provide this shared computing environment for the enterprise.  Interestingly enough, in order for SOA to work effectively, the same paradigm shift described earlier is also required here.  Applications supporting the entire enterprise cannot be funded and controlled by a silo.  Enterprise applications must be handled outside of the silo to be able to support every line of business.  Funding for enterprise services cannot be the responsibility of the silo as these will get lower priority compared to the silo's other initiatives.

Silos provide ownership for applications and shared enterprise applications also require ownership.  The owner needs to be responsible for the application so that decisions made and actions taken are for the benefit of all stakeholders and not just a specific business group.

Master data management (MDM) is one of these shared enterprise applications that requires the cooperation of all stake holders to be successful.  Master data management is an enterprise utility application that can benefit a wide cross section of the business and provide the controlled, shared environment for success.

Invest in Data Quality

Many organizations are not really aware of the quality of their data.  Quality issues normally arise as a result of business intelligence initiatives as they tend to highlight the situations where data is not reflecting what the business thinks it should.  Marketing campaigns having high levels of failed delivery highlight bad customer contact information, and other business processes that do not function as intended all highlight the same problem.

How can you recognize when you have trustworthy data quality?  If you can answer these questions with Yes, you most likely have trustworthy data quality:

  • Is my data accurate?   Do the values make sense?  Are the values meaningful?
  • Is my data valid?  Are the values in range?  Are the dates proper?
  • Is my data complete?  Are time periods missing?  Are mandatory fields empty?
  • Is my data consistent?  Do I get the same information from every source?
  • Is my information timely?  Can my data tell me what happened today, yesterday, last month?

If you are lucky enough to have a high level of data quality, then in order to maintain it you need data stewards who are responsible for dealing with any anomalies that may arise.  Data stewards respond to identified data issues and perform corrective actions whenever possible.  Common data issues that data stewards deal with are duplication, bad customer addresses, missing customer records and a host of other issues.

Improve Data Collection

Many data a issues area created at point of entry.  Data collection applications may not validate information enough, data collection methods may be inferior, and people entering data may not understand what's required or the importance of the data being collected.  When problems affecting data quality are identified, action should be taken to correct the problem as soon as possible.  If data was important enough to collect, it should be important enough to collect properly.

Control Data Sharing

Data being collected at one point and shared with other applications should be tracked and validated that the data is being sourced from the proper place and is fit for the intended purpose.  Data that is shared should also be sourced from the system of record.  If data is sourced from an application which is not the source, you risk some processing being done on that data that causes unwanted side effects.

Data replication is a fact of life.  Applications may need local copies of shared data to enable some functionality, or you may need to have a local copy to achieve the performance thresholds you required.  Replication is a valid process as long as the reasons for the replication are good enough to bypass using the central store, and the replication is properly documented so all downstream consumers of data are known.

Metadata and Documentation

One major aspect of data sharing is having the ability to locate, understand, and access the data.  A large client once estimated that 40% of all their IT projects was spent discovering data they already had.  Metadata, (data about data), is an important resource to supporting an enterprise data environment.  While creating and maintaining metadata is not usually high on the projects list of things to do, it can provide a wide variety of benefits like:

  • Intellectual property retention
  • Reduced IT costs by providing readily accessible information to support new development
  • Improved data quality by having clear definition of business rules
  • Better decision making by identifying appropriate sources for business purpose

Metadata creation and maintenance should be part of the project lifecycle, and like governance, have check points which must be satisfied to proceed with the project.  Metadata is most reliable when automation for the collection and maintenance of the metadata can be employed, but that only works for metadata of physical things, while high level definitions and descriptions require human intervention.

Monitoring your Data

Monitor your data so you can understand how it changes over time.  Monitor the quality of your data so problem sources can be identified earlier.  Monitoring your data can provide business insight.  If you are monitoring your MDM customer information you can see customer growth or decline, types of customers, and changes in customer behaviour.  Monitoring allows you to see trends in your data and your business so you can become more proactive and less reactive.

Data has a life cycle

Like tangible assets, your data has a life cycle and at some point it does not provide the intended value and should be retired. Data retention may be governed by legal obligations, by government legislation, or by business value.  Storage costs are significantly cheaper than before, but the hidden costs of managing all this historical data should not be over looked.

Identify when your data should be retired and either purge it or archive it for long term storage to allow your data environment to keep to a manageable size.

Enterprise data management is a program

If you've gotten this far and decided this is too much to tackle, you can relax as this is an evolutionary process that will take years.  Like any major initiative, trying to accomplish it in one big bang is rarely a good idea.  Enterprise data management is a program that is always active and growing, but it is a practice that cannot be ignored.  The scale of the problem may be different for every organization, but every  company will face the issues caused by unmanaged data.  Recognizing that enterprise data management is basically a corporate utility supporting all aspects of the business can help position it for easier adoption or expansion depending on the state of your enterprise data management practices.

Jan D. Svensson

Topics: Blog, MDM, Master Data Management, Data Management, enterprise data management

Abhishek Kamboj

Written by Abhishek Kamboj

Subscribe to Email Updates

Lists by Topic

see all

Posts by Topic

See all

Recent Posts