Please find us at: www.datavaluetalk.com

How much paper was used to deliver three copies of your glossy brochure in the same household? Every tried to estimate the impact on our environment of all the undeliverable direct mailers? How much fuel consumption can be reduced when orders  delivery can effectely be grouped together by geographic location?

Being more environmentally friendly is going to be a huge topic in the next few years. Effective data quality and customer data integration can go a long way into making a difference. A totally business to how we look at investing in these areas at this moment.

Service Oriented Architecture (SOA) has gained much popularity in the past years as the model for application development that uses groups of services that are orchestrated around business processes. SOA involves more than just the manner in which applications are deployed and interact with each other – it also involves how they interact with data. 

SOA

Service-enabling functions in the enterprise architecture allows for reuse and thus reduction in development and maintance cost. The functions running in the SOA stack we start to think of as reusable modules or components and the notion of an application gets elevated to something that can comprise one or more services. One of the fundamental things that makes this possible is acceptance of and adherence to standards—standards such as XML, SOAP, the Web services standards, etc.

When deploying data quality functions such as ‘name standardisation’, ‘address validation’ and ‘duplicate-matching’ as reusable services in the SOA stack then these can be integrated into any process that interacts with customer data. In the BMPS layer data quality related processes like ‘Insert Party’ or ‘Blacklist check’ can be assembled and made available to every process and application in the SOA stack. Data quality deployed in this manner can provider greater flexibility, control and data consistency.

Last week I made a post on a Gartner Survey that showed increasing SaaS momentum in the economic down-turn. Well, the industry analyst seem to be tumbling all over each other to report on the accellerating pace of SaaS adoption, especially in relation to the current situation.

A couple of days ago IDC published a press release on cloud computing and how it is reshaping the IT marketplace. 

Some major points I think are very interesting:

• Cloud computing is “crossing the chasm” and entering a period of widespread adoption
• IDC expects the cloud adoption trend to be amplified by the current financial crisis
• IT cloud services to grow almost threefold, reaching $42 billion by 2012
• More importantly, cloud computing will account for 1/3 of IT growth in 5 years

Just before the upcoming elections in the United States on November 4th many voters have fallen victim to incorrect voters registration. It might well involve over hundred thousands of voters.

The Help America Vote Act was passed in 2002 in order to prevent voter registration and voting counting problems which happened with the presidential elections in 2000. One of the measures was to create centralized databases, which allow voters to check their registrations. In rolling up and centralizing local databases to state databases many mishaps have occured. Resulting in thousands of voters who are now not eligible to vote. These people have to reestablish their eligibility to vote in the coming weeks.

Click here for the full article in the Washington Post.

Jill Dyché, an expert on Customer Data Integration, was asked about the most common mistakes and obstacles when organizations want to implement CDI. For the complete article please follow this link. She came up with 7 very usefull tips:

1) Anticipate the saboteurs. With CDI there will be people in your organization claiming that “we already do that with our data warehouse/ODS/CRM system/toothpicks-and-glue.” Anticipate their arguments and educate them, offering deliberate examples of why they’re full of it. Nicely, of course.

2) Articulate more than one business problem that CDI can solve. You want to position your CDI effort as an ongoing program that can enable different business needs. If you position CDI as a one-trick pony (for instance, it can handle identity resolution), you’ll only be able to ride that horse for so long…

3) Know your functional requirements. It never ceases to amaze me how quickly companies want to enlist vendors. “We’ve gotten everyone in a room and we know what we need,” they explain as they cherry-pick brochures off of vendor websites and schedule proofs of concept (POC). Not so fast! At best, this is a well-intentioned attempt to fill in the blanks by talking to smart vendors. At worst, it’s tire-kicking in a vacuum. Unlike with business intelligence (BI), master data management (MDM) requires a huge and measured focus on functional requirements. Until you have your list in hand, you won’t be able to give the vendors what they need to deliver the goods.

4) Don’t underestimate the need for savvy development skills. This isn’t your father’s data warehouse. Many IT environments are so accustomed to buying off-the-shelf applications that they lack the development skills necessary to configure and maintain an MDM solution. Not that it’s a science project, but it’s not a packaged app either. If your business is complex, your MDM solution will be, too.

5) Don’t prematurely launch data governance. Not to be an alarmist; I’ve never yelled “Fire!” in a crowded theater. (Though I have yelled “Fire!” twice, both times during Terms of Endearment.) But you can rush data governance, and the results can be disastrous and long-lasting. If someone asks you to come to a council meeting, and they haven’t defined data governance for you, duck and cover.

6) Don’t just assume the enterprise application vendors will fit the bill. Just because you already have a [fill in the blank] license doesn’t mean that’s the right MDM product to suit your needs. See number three, above, or call me.

7) If you’re not sure where to start, convene presumptive stakeholders for an executive workshop that defines MDM, explains the pros and cons, profiles the key vendor products, and foreshadows the “gotchas.” A good consulting company can deliver such a workshop and follow it up with an Observations Report that highlights key conversations and questions and recommends tactics for moving forward.

Already in 2006 Jill Dyché and Evan Levy published “Ten Mistakes to Avoid”.

Jeff Kelly from SearchDataManagerment.com writes about the momentum of SaaS based data quality and integration tools. With quotes from a survey done by research firm Gartner he writes about how “the on demand model is beginning to gain foothold in the industry”.

A few remarkable facts from the Gartner survey:

• 28% of companies have deployed SaaS-based data integration tools
• 24% have SaaS-based data quality tools
• Most companies use a mix of on-premise, outsourced and SaaS-based technology
• SaaS-based software, at this point, often trails its on-premise counterparts
• Gartner advice: “Plan SaaS deployment enterprise-wide….to guard against siloed investments”

Enterprise-wide data quality project requires custom implementation and customisation of the tools to support the specific business processes. This maybe difficult to achieve in the data quality SaaS offerings that are available at this time. On  the other hand many data quality functions like rapid addressing, data enrichment or even deduplication happen all across the enterprise and may be well served with on demand offerings.

The current economic crisis will force many organisations to reduce IT expenses and SaaS based data quality will have increasing appeal. In most organisations a mix between on demand services and on premise implementation will be the best model.

Planet Google (Book)

A new book by Randall Stross is out (6th October 2008); it’s called “Planet Google” and covers Google’s beginnings but also their latest actions and troubles. Some quotes from the book triggered me. I hope Google Health will be addressed also.

Google has been determined to “organize the world’s information and make it universally accessible and useful” since June 1999 … Until then, Google had used a modest statement of company mission that the founders had hastily put on Google’s Web site at the time the site was launched: “To make it easier to find high-quality information on the web.”

“In looking back, [Google’s Brian Rakowski] marveled at the public’s focus on the scanning for matching ads [in Gmail], while ignoring much more serious privacy issues, such as the increased risk that e-mail on centralized servers could easily be subpoenaed or personal information in e-mail messages could be shared with advertisers. “Nobody was talking about those privacy issues and the few things we were doing to protect users’ privacy,” he said.”

HI Survey Results

Looking at the data can give you deeper insight in your data processes. Not only the processes in theory, but in practice too. How? By focusing on the exceptions!

For example, do you think the field ‘social security number’ is reliable because it is mandatory and uses a modulo 11 check digit? Maybe this field is filled with a fake-value like ‘111222333′ , or the social security number of the agent who entered the data. A frequency check on the values in this field will identify values that occur more frequently than others.

 Why do people do that, cheat? One reason may be that staff at the front desk want to help their customer, or sell something, and the customer doesn’t know their social security number at that moment.

 The same applies to all other fields.  Address fields mandatory, but the customer doesn’t want to give their address ? Fill in the address of the office. Customer deceased? Put the text (deceased)  after their surname. Ex-directory phone number? Use the email field for entering the text ‘ex-directory’.

 Get to know your processes by analyzing your data!

As a recent Human Inference survey shows, data quality processes appear to be the biggest challenge for organizations.  Although this may seem an obvious survey result for many, the real question here of is: What are data quality processes?

Many of you will probably start thinking along the lines of data ownership, database access and hierarchy. But data quality processes are far more essential.  First of all, you have to know who you are dealing with.The following video is self-explaining: