Brain Vibe

social media marketing muses to stay engaged

December 2, 2009 • 12:23 am 2

Data Governance More Than Ownership

When kicking off data management initiatives a large and key component is establishing the data stewards that represent the data that is collected, managed, and leveraged in business intelligence.  By having these data stewards, and subsequently a data management committee, companies feel safe that the proper data governance practices are going to be put in place.  Not so.  Ownership (=Stewardship)  does not equate to governance.

Many factors contribute to governance and business boundaries can quickly be broken down if you approach governance in business silos.  As you walk through your process of data collection you’ll quickly find that what is considered the preferred source of data may not be generated by the team that determines what should stay, what should be modified, and what should go.  In fact, depending on how you view the data, conflicts arise as to what is considered accurate, appropriate, of the contributing factor in decision and business point of view.

This is something I’ve run into recently when building a business intelligence solution for web analytics.  Even within my own department of advertising executives, views of what transactional data should be considered the record of source is up for grabs depending on who is the recipient of the information and how it is used.  Levels of accuracy vary depending on when data is needed, how it may be used for marketing optimizations, or if it will be used to actualize spending for billing.  Throw into the mix that data feeds coming from vendors are constantly changing as they actualize transactions over the course of days, weeks, and even months, and finding the truth in the data becomes a challenge that defies religious opinion on the subject.

Sorting through the challenges of governance to determine what makes data reliable requires looking at a variety of factors and allowing for multiple views and uses.

  • Reliability of source
  • Time of collection
  • Actualization
  • Business process affected/use of data in decisions
  • Degree of accuracy required

If you will notice, I do not include ownership.  This is the artificial governance.  Ownership in establishing governance only serves to create a framework around the above factors that creates credibility.  Ownership, and then the transformation to stewardship, serves to continuously monitor, enforce, and improve governance around data needs.

Start your data management off on the right foot, don’t confuse ownership with governance.

Filed under: business intelligence, data quality , , , ,

August 10, 2009 • 10:11 pm 0

Ensuring quality data from service providers

For those of us that have lived, eaten, and slept with data quality and data management it is hard to fathom that there are still pockets of those that have yet to define a solid foundation of data quality and data management best practices.  It is even harder still to take a step (leap) back into the roots of how data quality and data management issues all began.  Well, let me tell you, those pockets of organizations are alive and well in the most unlikely places – those companies that are providing data.

To be fair, there are some amazing companies out there that provide information and data that we use to improve and enhance our own data or take to analyze independently.  They may not be perfect (no one is!).  Though, they have defined themselves as servicing organizations with “better quality data” and stand by it with best practices of their own.  But, as enterprise organizations and even mid-sized companies have jumped on the band wagon and adopted sophisticated processes, solutions, and people that are dedicated to better information, there are still a significant number of services providers that lack the skills, tools, and practices that would ensure reliable information to measure our performance, understand our market, and take advantage of new opportunities.

At the end of the day, the data and information we source needs to be reliable.  It is important to guard yourself when both contracting with service providers and when you receive data.  Simply relying on the fact that the data is of high quality when you receive it is not good enough.  You need to be vigilant during the sourcing of providers as well as clearly defining how you can ensure what you received is what you paid for.  Here are some things to consider and ask when working with data providers:

  • How do they collect their information?
  • How do they verify that the information is valid?  What process, sources, and analysis is used?
  • Are they providing data to other customers for the same purpose you need the information for?  How many/what portion?
  • What is their repeat business rate?  Who are their top customers?
  • What purposes are their customers using their data?
  • What do they do to verify and validate your data prior to providing it to you?
  • What do they do to verify that the data they are providing is complete?
  • What guarantees do they or will they provide that the data meets your specifications and quality standards?
  • What is required on your end to validate that the data is accurate and reliable?
  • If you are purchasing tracking data (real time/period feeds), what initial and regular testing processes used to verify proper data transfers?
  • What is required on your end to ensure the data transfer is working initially and ongoing?

What have you done to ensure data from service providers is what you want?

Filed under: data quality , , , , ,

April 13, 2009 • 1:00 pm 0

Archiving Strategy: Data Relevance

We often think of the relavence of data when we want to include or exclude it from analysis or process.  However, are you thinking about relavence as part of your data quality effort?

Just as you focus data quality efforts to clean existing information, there are invariably records that can’t be cleansed or enhanced.  They have no value in either business analytics or business process.  They are noise, similar to the noise you have when there is bad data.  To save and maintain them in your database can affect your ability to accurately analyze information, continue to deflate confidence in data, and if a significant percentage of your database, will cause problems in performance and added maintenance.  Developing an archival strategy as part of your data quality practice is a significant component that should not be overlooked.

Benefits of Data Relevance

  • Trust in data
  • Enables process
  • Accuracy of analysis
  • Supports decisions
  • Database optimization

It can be tempting to simply delete records from your databases.  Though, this can have a detrimental affect due to data dependencies within your databases as well as causing non-compliance in regulated environments.  Instead, it is best to formulate a strategy that flags non-relevant data removing or suppressing it from user interfaces and analytics.

Components of Archiving Strategy

  • Data decay rates – Attributes of records that loose relevance over time.  This component is a good guide on the frequency at which you will focus cleansing efforts.  It also provides an indicator on when data is approaching a horizon when a record will lose its relevance.  Age of the data and activity related to a record, even if a record is complete, can signify whether the data is relavant and open to archiving.
  • Minimum requirements of record viability – Records should continually be assessed to determine if they meet the minimum standards of use.  Failure to meet minimum requirements is a leading indicator that the record is a candidate for archiving.
  • Relevance of record to analysis, process, decisions – If a record is not going to be used in analysis, process, or decision making, there is not need to keep it in use.  This may be the case if processes have been optimized and certain information is no longer needed.  Or, it could be that it was a candidate for archiving due to decay rates and minimum data requirements.  Additionally, relavance may be determined when integrating systems where old records with old transaction history is not relevant to the existing or new business.
  • Regulatory compliance – In highly regulated environments like health care, there are standards on what you can and cannot remove.  Records may not be useful in existing process, analysis, and decision making, but might be required in certification or other compliance related activities.  Archiving ensures that information is not deleted from primary systems.  Although, you may have to provide a mechanism that provides adequate access to data for compliance.

An archiving strategy is a critical component of data quality best practices.  It will continually help you focus on improving and refining your data quality projects as well as thinking strategically about how you use and manage your data on a daily basis.  Establish an archiving strategy at the forefront of your data quality initiatives and you start your efforts off on the right foot.

Reblog this post [with Zemanta]

Filed under: Uncategorized , , , , , , , , , , ,

April 4, 2009 • 1:54 pm 0

Stuck in First Gear

porscheBig investments were made in recent years in IT.  IBM, Oracle/Siebel and SAP lead the market and were successful not only with the multi-national enterprise companies but, also with mid-sized companies.  There are a lot of companies out there that have purchased application and data management/data warehouse solutions only to find themselves using a portion of what it could do.  It’s like driving a Porcshe in first gear.

There are some fundamental reasons for this, outside of the fact that companies may feel it is the fault of their sales execute selling them the wrong bill of goods.  IT will blame the business for not knowing what it wants.  The business will blame IT for not getting it.  Doesn’t really matter, there is plenty of blame to go around.  What matters is that now you have a solution that isn’t giving you the benefits that it really could and should be.

Maybe I’m a bit biased since I’m the data chick.  Well, more than a bit.  Regardless, I think that from a data management perspective, companies are failing.  The maniacal focus on process efficiency has drowned out the fact that process runs on data and feeds data.  This focus has put data in the back seat too long and now when we need it to better understand our customers, our business, and make decisions, it is sorely lacking.  Our data lacks unity, structure, definition, and most of all purpose.  Companies simply cannot leverage their information except at very basic levels.  When things are good, this may be okay.  When things are bad, this is a real problem.

What makes this even more sad, is that companies are looking to spend more money on applications and data infrastructure to ‘fix’ the problem.  The promise of the new model and more sophisticated bells and whistles that will solve anything you throw at it is just marketing.  Until you can understand and control what you already have under your hood, getting something bigger, better, and shinier isn’t going to help anymore than it does now.  So, there was no ROI on existing purchases and there won’t be any ROI on new purchases.

There are two things companies need to do to make the investments in enterprise solutions worthwhile:

  • Clean-up the back-end data management practice so that it is fluid with business process and application usage.
  • Have a clear data management strategy for new applications that is fluid and scalable outside of application databases.

Your company may already be embarking on SOA or MDM projects.  But, have you looked at how these new practices will support applications outside of changing the oil?  Can the data drive process?

Today, applications are bogged down because data is treated as something to put in the trunk and horde.  Until data is thought of as fuel, you’re IT investments will stay in 1st gear and never get to 6th.  Now how fun is that?

Reblog this post [with Zemanta]

Filed under: Uncategorized , , , , , , , ,

March 26, 2009 • 7:09 pm 2

How to Measure the Business Impact of Data Quality

So, you want to invest in data quality but you need to prove ROI before you get the resources. Intuitively you know that data quality is impacting your business. How to measure that to make a case is the test.

Many businesses focus on data elements that are easy to see and understand like company and contact information.  However, as obvious as some of these elements may be, they don’t always lead to the highest bang for the buck.  Data elements have priority levels within processes depending on the desired business outcome.  In addition, data elements have dependencies outside of how the information comes into the system.  You need to take this into account as you conduct your business analysis and map your data across your business processes.

During business analysis it pays to establish a foundation that validates recommendations and shows ROI through case studies.  You can do this through data analysis and pilot programs.  Data analysis can be applied through meta data segmentation within processes where you look at the existing state of the data.  You can also improve portions of the data and perform the segmentation and analysis.

These steps will prepare your case but will also help establish dashboards to allocate resources for future projects.

1) Identify the processes you think are most impacted by poor data quality . The processes should be tied into key business functions. For instance, in marketing you may want to look at lead qualification and management. Processes that are well defined and have a tangible link to businesses objectives work best as they are most likely mature and revenue has been tied to them.
2) Pinpoint smoking guns in the processes.  There are bound to be several points in a process that are key indicators of success where data quality has negatively impacted the outcome.  Your business analysis will or should show this clearly.  These smoking guns should be called out clearly in the processes.  What you should determine is which data elements are impacting the most and can be easily focused on or addressed.
3) Select data quality issues that you can segment the process into influence tracks.  This step is critical to measurement.  You need to dissect the process to create scenarios of what good vs. bad looks like in process outcomes.  In the lead management process suggested earlier, it could be the point where you would qualify a lead to move into the sales pipeline.  
4) Measure performance success with good quality vs. poor quality data.  At this stage you should be able to run an analysis that shows the difference in process outcomes and performance when you run scenarios between good quality data and poor quality data.  

The real benefit is that at this stage you’ve provided the dashboard to measure improvements to the business.  Rather than wait until the data quality projects are completed, this provides the foundation for predicting where you will get the most impact from your investments. Instead of focusing solely on metrics that measure the completeness, accuracy, and uniqueness of records, you can focus on how these metrics within processes influence business outcomes.  Now you have a case for linking data quality with ROI.

Filed under: data quality , , , , ,

News

Intelligent Metrix now integrated with Brain Vibe. Strategy driven by intelligence. Search for articles on business intelligence, CRM, and data quality.

Top B2B Blogs

Topics

Blog Archive

Linking

Bookmark and Share
Featured blogger at Smart Data Collective

 

March 2010
M T W T F S S
« Dec    
1234567
891011121314
15161718192021
22232425262728
293031  

Linking

Subscribe to Brain Vibe by Email

Follow Me On Twitter

Add to Technorati Favorites

Bookmark and Share

 Subscribe in a reader

Networks

Business Blogs - Blog Catalog Blog Directory