The Data Intelligence Gap: Part One

There is a huge chasm in many corporations today, one that hurts companies by keeping them from revenue, more profit, and better operating efficiency. The gap, of course, lies in corporate information.

On one side of the gap lies corporate data, which may contain anything from unintelligible garbage to very valuable data. However, it is often very challenging to identify the difference. On the other side of the chasm are business users, ever needing stronger corporate intelligence, longing for ways to stimulate corporate growth and improve efficiency. On this side of the chasm, they know what information is needed to make crucial decisions, but are unsure if the data exists to produce accurate information.

Data needs standardization, organization and intelligence, in order to provide for the business.

Companies often find themselves in this position because rapid corporate growth tends to have a negative impact on data quality. As the company grows and expands new systems are put in place and new data silos form. During rapid growth, corporations rarely consider the impact of data beyond the scope of the current silo. Time marches on and the usefulness of data decays. Employee attrition leads to less and less corporate knowledge about the data, and a wider gap.

So, exactly what is it that the business needs to know that the data can’t provide? Here are some examples:

What the Business Wants to Know	Data needed	What’s inhibiting peak efficiency
Can I lower my inventory costs and purchase prices? Can I get discounts on high volume items purchased?	Reliable inventory data.	Multiple ERP and SCM systems. Duplicate part numbers. Duplicate inventory items. No standardization on parts descriptions and numbers. Global data existing in different code pages and languages.
Are my marketing programs effective? Am I giving customers and prospects every opportunity to love our company?	Customer attrition rates. Results of marketing programs.	Typos. Lack of standardization of name and address. Multiple CRM systems. Many countries and systems.
Are any customers or prospects “bad guys”? Are we complying with all international laws?	Reliable customer data for comparison to “watch” lists.	Lack of standards. Ability to match names that may have slight variations against watch lists. Missing values.
Am I driving the company in the right direction?	Reliable business metrics. Financial trends.	Extra effort and time needed to compile sales and finance data – time to cross-check results.
Is the company we’re buying worth it?	Fast comprehension of the reliability of the information provided by the seller.	Ability to quickly check the accuracy of the data, especially the customer lists, inventory level accuracy, financial metrics, and the existence of “bad guys” in the data.

Again, these are some of the many reasons where data lacks intelligence and can’t provide for the needs of the corporation. It is across this divide that data quality vendors must set up bridges... and it is this chasm that data governance teams must cross.

We’ll cover that in part two of this story. I’ll cover what kind of solutions and processes help the data governance team cross the great data divide and bring data intelligence to their organizations.

Forget the Data. Eat the Ice Cream.

It’s summer and time for vacations. Even so, it’s difficult for a data-centric guy like me to shut off thoughts of information quality, even during times of rest and relaxation.
Case in point, my family and I just took a road trip from Boston to Burlington, VT to visit the shores of Lake Champlain. We loaded up the mini-van and headed north. Along the way, you drive along beautiful RT 89, which winds its way through the green mountains and past the capital - Montpelier.
No trip to western Vermont is complete without a trip to the Ben and Jerry’s ice cream manufacturing plant in Waterbury. They offer a tour of the plant and serve up a sample of the freshly made flavor of the day at the end. The kids were very excited.
However, when I see a manufacturing process, my mind immediately turns to data. As the tour guide spouted off statistics about how much of any given ingredient they use, and which flavor was the most popular (Cherry Garcia), my thoughts turned to the trustworthiness of the data behind it. I wanted him to back it up by telling me what ERP system they used and what data quality processes were in place to ensure the utmost accuracy in the manufacturing process. Inside, I wondered if they had the data to negotiate properly with the ingredients vendors and if they really knew how many heath bars, for example, they were buying across all of their manufacturing plants. Just having the clean data and accurate metrics around their purchasing processes could save them thousands and thousands of dollars.
The tour guide talked about a Jack Daniels flavored ice cream that was now in the “flavor graveyard” mostly because the main ingredient was disappearing from the production floor. I thought about inventory controls and processes that could be put in place to stop employee pilfering.
It went on and on. The psychosis continued until my daughter exclaimed “Dad. This is the coolest thing ever! That’s how they make Chunky Monkey!” She was right. It was perhaps the coolest thing ever to see how they made something we use nearly every day. It was cool to take a peak inside the corporate culture of Ben and Jerry’s. It popped me back into reality.
Take your vacation this year, but remember that life isn’t only about the data. Remember to eat the ice cream and enjoy.

The Soft Costs of Information Quality

Choosing data quality technology simply on price could mean that you end up paying far more than you need to, thanks to the huge differences in how the products solve the problems. While your instinct may tell you to focus solely on the price of your data quality tool, your big costs come in less visible areas – like time to implement, re-usability, time spend preprocessing data so that it reads into the tool, performance and overall learning curve.

As if it wasn’t confusing enough for the technology buyer having to choose between a desktop and enterprise-class technology, local and global solutions, or built-in solution vs. universal architecture, now you have to work out soft costs too. But you need to know that there are some huge differences in the way the technologies are implemented and work day-to-day, and those differences will impact your soft costs.

So just what should you look for to limit soft costs when selecting an information quality solution? Here are a few suggestions:

• Does the data quality solution understand data at the field level only or can it see the big picture? For example, can you pass it an address that’s a blob of text, or do you need to pass it individual first name, last name, address, city, state, postal code lines. Importance: If the data is misfielded, you’ll have a LOT of work to do to get it ready for the field level solution.
• On a similar note, what is the approach to multi-country data? Is there an easy way to pre-process mixed global data or is it a manual process? Importance: If the data has mixed country of origin, again you’ll have to do a lot of preprocessing work to do to get it ready.
• What is the solution’s approach to complex records like “John and Diane Cougar Mellencamp DBA John Cougar”? Does the solution have the intelligence to understand all of those people in a record or do I have to post-process this name?
• Despite the look of the user interface, is the product a real application or is it a development environment? Importance: In a real application, an error will be indicated if you pass in some wild and crazy data. In a development environment, even slight data quirks will cause nothing to run and just getting the application to run can be very time consuming and wasteful.
• How hard is it to build a process? As a user you’ll need to know how to build an entire end-to-end process with the product. During proof of concept, the data quality vendor may hide that from you. Importance: Whether you’re using it on one project, or across many projects, you’re eventually going to want to build or modify a process. You should know up-front how hard this is. It shouldn’t be a mystery, and you need to follow this during the proof-of-concept.
• Are web services the only real-time implementation strategy? Importance: Compared to a scalable application server, web services can be slow and actually add costs to the implementation.
• Does the application actually use its own address correction worldwide or a third party solution? Importance: Understanding how the application solves certain problems will let you understand how much support you’ll get from the company. If something breaks, it’s easier for the program’s originator to fix it. A company using a lot of third party applications may have challenges with this.
• Does the application have different ways to find duplicates? Importance: During a complex clean-up, you may want to dedupe your records based on, say e-mail and name for the first pass. But what about the records where your e-mail isn’t populated? For those records, you’ll need to go back and use other attributes to match. The ability to multi-match allows you to achieve cleaner, more efficient data by using whatever attributes are best in your specific data.

I could go on. The point is – there are many technical, in-the-weeds differences between vendors, and those differences have a BIG impact on your ability to deliver information quality. The best way to understand a data quality vendor’s solution is to look over their shoulder during the proof-of-concept. Ask questions. Challenge the steps needed to cleanse your data. Diligence today will save you from having to buy Excedrin tomorrow.

Data Quality Events – Powerful and Cozy

For those of you who enjoy hobnobbing with the information quality community, I have a couple of recommendations for you. These events are your chance to rub elbows with different factions of the community. In the case of these events, the crowds are small but the information is powerful.

MIT Information Quality Symposium
We’re a couple of weeks away from the MIT Information Quality Symposium in Boston. I’ll be sharing the podium with a couple of other data quality vendors in delivering a presentation this year. I’m really looking forward to it.
Dr. Wang and his cohorts from MIT fill a certain niche in information quality with these gatherings. Rather than a heavily-sponsored, high pressure selling event, this one really focuses on concepts and the study of information quality. There are presenters from all over the globe, some who have developed thought-provoking theories on information quality, and others who just want to share the results of a completed information quality project. The majority of the presentations offer smart ways of dissecting and tackling data quality problems that aren’t so much tied to vendor solutions as they are processes and people.
My presentation this year will discuss the connections between the rate at which a company grows and the degree of poor information in the organization. While a company may have a strong desire to own their market, they may wind up owning chaos and disorder instead, in the form of disparate data. It’s up to data quality vendors to provide solutions to help high-growth companies defeat chaos and regain ownership of their companies.
If you decide to come to the MIT event, please come by the vendor session and introduce yourself.


Information and Data Quality Conference
One event that I’m regrettably going to miss this year is Larry English’s Information and Data Quality Conference (IDQ) taking place September 22-25, in San Antonio, Texas. I’ve been to Larry’s conferences in past years and have always had a great time. What struck me, at least in past years, was the fact that most of the people who went to the IDQ conference really “got it” in terms of the data quality issue. Most of the people I’ve talked with were looking for sharing advice on taking what they knew as the truth – that information quality is an important business asset – and making believers out of the rest of their organizations. Larry and the speakers at that conference will definitely make a believer out of you and send you out into the world to proclaim the information quality gospel. Hallelujah!

Thanks
On another topic, I’d like to thank Vince McBurney for the kind words in his blog last week. Vince runs a blog covering IBM Information Server. In his latest installment, Vince has a very good analysis of the new Gartner Magic Quadrant on data quality. Thanks for the mention, Vince.

Get Smart about Your Data Quality Projects

With all due respect to Agent Maxwell Smart, there is a mini battle between good and evil, CONTROL and KAOS, happening in many busy, fast-growing corporations. It is, of course, with information quality. Faster growing companies are more vulnerable to chaos because by opening up new national and international divisions, expanding through acquisition, manufacturing offshore, and doing all the other things that an aggressive company does, it leads to more misalignment and more chaotic data.

While a company may have a strong desire to “own the world” or at least their market, they may wind up owning chaos and disorder instead - in the form of disparate data. The challenges include:

• trying to reconcile technical data quality issues, such as different code pages like ASCII, Unicode and EBCDIC
• dealing with different data quality processes across your organization, each that that deliver different results
• being able to cleanse data from various platforms and applications
• dealing with global data, including local languages and nuances

Agent 99: Sometime I wish you were just an ordinary businessman.
Maxwell Smart: Well, 99, we are what we are. I'm a secret agent, trained to be cold, vicious, and savage... but not enough to be a businessman.

In an aggressive company, as your sphere of influence increases, it’s harder to gather key intelligence. How much did we sell yesterday? What’s the sales pipeline? What do we have in inventory worldwide? Since many company assets are tied to data, it’s hard to own your own company assets if they are a jumble.

Not only are decision-making metrics lost, but opportunity for efficiency is lost. With poor data, you may not be able to reach customers effectively. You may be paying too much to suppliers by not understanding your worldwide buying power. You may be driving your own employees away from innovations, as users begin to avoid new applications because of data.

KAOS Agent: Look, I'm a sportsman. I'll let you choose the way you want to die.
Maxwell Smart: All right, how about old age?

So, it’s up to data quality vendors to provide solutions to help high-growth companies “get smart” and defeat chaos (kaos) to regain ownership of their companies. They can do it with smart data-centric consulting services that help bring together business and IT. They can do it with technology that is easy to use and powerful enough to tackle even the toughest data quality problems. Finally, they can do it with a great team of people, working together to solve data issues.

Agent 99: Oh Max, you're so brave. You're going to get a medal for this.
Maxwell Smart: There's something more important than medals, 99.
Agent 99: What?
Maxwell Smart: It's after six. I get overtime.

Probabilistic Matching: Part Two

Matching algorithms, the functions that allow data quality tools to determine duplicate records and create households, are always a hot topic in the data quality community. In a previous installment of the Data Governance and Data Quality Insider, I wrote about the folly of probabilistic matching and its inability to precisely tune match results.

To recap, decisions for matching records together with probabilistic matchers are based on three things: 1) statistical analysis of the data; 2) a complicated mathematical formula, and; 3) and a “loose” or “tight” control setting. Statistical analysis is important because under probabilistic matching, data that is more unique in your data set has more weight in determining a pass/fail on the match. In other words, if you have a lot of ‘Smith’s in your database, Smith becomes a less important matching criterion for that record. If the record has a unique last name like ‘Afinogenova’ that’ll carry more weight in determining the match.

The trouble comes when you don’t like the way records are being matched. Your main course of action is to turn the dial on the loose/tight control to see if you can get the records to match without affecting record matching elsewhere in the process. Little provision is made for precise control of what records match and what records don’t. Always, there is some degree of inaccuracy in the match.

In other forms of matching, like deterministic matching and rules-based matching, you can very precisely control which records come together and which ones don’t. If something isn’t matching properly, you can make a rule for it. The rules are easy to understand. It’s also very easy to perform forensics on the matching and figure out why two records matched, and that comes in handy should you ever have to explain to anyone exactly why you deduped any given record.

But there is another major folly of probabilistic matching – namely performance. Remember, probabilistic matching relies heavily on statistical analysis of your data. It wants to know how many instances of “John” and “Main Street” are in your data before it can determine if there’s a match.

Consider for a moment a real time implementation, where records are entering the matching system, say once per second. The solution is trying to determine if the new record is almost like a record you already have in your database. For every record entering the system, shouldn’t the solution re-run statistics on the entire data set for the most accurate results? After all, the last new record you accepted into your database is going to change the stats, right? With medium-sized data sets, that’s going to take some time and some significant hardware to accomplish. With large sets of data, forget it.

Many vendors who tout their probabilistic matching secretly have work-arounds for real time matching performance issues. They recommend that you don’t update the statistics for every single new record. Depending on the real-time volumes, you might update statistics nightly or say every 100 records. But it’s safe to say that real time performance is something you’re going to have to deal with if you go with a probabilistic data quality solution.

Better yet, you can stay away from probabilistic matching and take a much less complicated and much more accurate approach – using time-tested pre-built business rules supplemented with your own unique business rules to precisely determine matches.

Data Profiling and Big Brown

Big Brown is positioned to win the third leg of the Triple Crown this weekend. In many ways picking a winner for a big thoroughbred race is similar to planning for a data quality project. Now, stay with me on this one.

When making decision on projects, we need statistics and analysis. With horse racing, we have a nice report that is already compiled for us called the daily racing form. It contains just about all the analysis we need to make a decision. With data intensive projects, you’ve got to do the analysis up front in order to win. We use data profiling tools to gather a wide array of metrics in order to make reasonable decisions. Like in our daily racing form, we look for anomalies, trends, and ways to cash in.

In data governance project planning, where there are company-wide projects abound, we may even have the opportunity to pick the projects that will deliver the highest return on investment. It’s similar to picking a winner at 10:1 odds. We may decide to bet our strategy on a big winner and when that horse comes in, we’ll win big for our company.

Now needless to say, neither the daily racing form nor the results of data profiling are completely infallible. For example, Big Brown’s quarter crack in his hoof is something that doesn’t show up in the data. Will it play a factor? Does newcomer Casino Drive, for whom there is very little data available, have a chance to disrupt our Big Brown project? In data intensive projects, we must communicate, bring in business users to understand processes, study and prepare contingency plans in order to mitigate risks from the unknown.

So, Big Brown is positioned to win the Triple Crown this weekend. Are you positioned to win on your next data intensive IT project? You can better your chances by using the daily racing form for data governance – a data profiling tool.