One of the first things that I do when I write an installment of “Eye” is reflect on what I wrote last year. When I did it this time, I was impressed by how consistently great this conference is. The weather in San Diego is unfailingly warm and sunny. Seaforth marina, with a fresh fleet of sail boats for rent is steps from the hotel.
Just beyond the marina, is a park that is great for jogging, playing the didgeridoo, or just watching the world go by. There are good restaurants in the neighborhood, and the hotel, the Manchester Grand Hyatt, is top notch. Of course, the main draw was TDWI’s summer conference. However, like our perceptions of the physical universe, a significant portion of the content of TDWI conferences does change, and this meeting was no different. In fact, TDWI’s ability to identify and integrate new thinking about the world of BI and data warehousing has been the hallmark of its conferences, and I was looking forward to this new installment.
I arrived in time to drop in on the TDWI partner members welcoming reception. Several friends from Connect: The Knowledge Network were there and we had a good chat. From their perspective, the BI job market continues to grow. Connect was adding to staff, and their clients had a number of open positions for consultants and full time employees. That was reassuring for BI professionals in this changing world.
Frank Buytendijk’s keynote, “The Future of Performance Management: Managing the Heterogeneous World,” got the juices flowing right away with his challenge to change our thinking about BI. He observed that this was a “2.0” world, which, along with wife 2.0, was characterized by co this and co that, i.e. co-marketing, co-development, co-research, and co-creation.
Buytendijk asserted that hierarchies were so 20th century and that networks were new century thinking, especially in performance management. As a thought-leader in PM, he stressed that we build a business model and the measures to manage it based on horizontal interactions with customers and those involved with jointly producing goods and services. He also observed that SLAs (service level agreements) led to inflexibility and an “us versus them” mentality. Buytendijk exhorted us to share BI with our customers and develop reciprocal metrics between regulators, the community, investors, and employees. He concluded with a reminder that measurement drives behavior and reciprocal metrics support collaboration.
After the keynote I sat in on Thomas Redman’s advanced course in data governance. Redman started out with a brief recognition that data governance programs are usually understaffed and lack necessary political clout. He then gave us three tests that were designed to assess the effectiveness of data governance, or lack there-of, in our organizations. The “care and feeding” test asked us if we had the right data, whether people could access them, whether people trusted them, and whether reasonable measures were in place to keep them safe. The other two tests involved whether we used data effectively in running the business, and whether we used them to change our organizations. The wording on these tests was pretty basic and easy for the average business user to understand. Even if we had not gone any further, Redman had given us good ammunition to get stalled governance programs restarted. However, he went on to give a number of looks at how to address data quality issues and the politics surrounding data management.
At the mid-day break, TDWI hosted a luncheon program that recognized best practices in a number of different areas. After the formal part of the program was over, we had a chance to talk one-on-one with different winners. I chatted with one of them that was involved with the City of Richmond, VA Police Department. The department had used a combination of predictive analysis from SPSS, BI technology from Information Builders, and ESRI GIS (geographical information systems) capabilities to reduce major crime by between 19 and 21% over the past two years. This story warmed the cockles of my heart, since in a past life I had been a product manager for several different vendors. I often found myself being asked by organizations that had two or three of these technologies which one they should select as the standard. All too often, the tools addressed different needs and, in my humble opinion, these organizations were all-too-often hell bent on decreasing the value of their data through restricting its access and use to a single tool. Stupid! Stupid! Stupid! In this case, Information Builders and ESRI had a strategic alliance that made the marriage of BI and GIS a bit easier to grasp. I was glad to see that the Richmond Police Department not only “got it,” but that they got recognized for it by the judges for TDWI’s Best Practices competition. Yea! And oh by-the-way, reduced crime was the result!
By evening, I was ready to kick back so I stopped by the Baja Lounge, a hospitality suite hosted by DataFlux. San Diego is a great town for south-of-the border food, drink, and entertainment, and the folks at DataFlux made the best of it. They were munificent hosts. The food from the tapas bar was exquisite in taste and wide ranging in variety. I met Don Julio 1942, a most magnificent elixir of blue agave, at the tequila bar, and I had the opportunity to chat with Evan Levy and Jill Dyche’ who were signing their new book, Customer Data Integration: Reaching a Single Version of the Truth. I also ran into and reconnected with a number of old friends, all while swaddled in the sounds of a great contemporary Latin jazz duo, MalaMana made up of Junanito Tidwell and Fredi Kunze.
I had found Frank Buytendijk’s keynote thought-provoking and was looking forward to his follow-up session, “Advanced Performance Management—Organizational Behavior and Strategic Alignment.” Buytendijk spent the first part of the morning expanding on his keynote which focused on horizontal as well a vertical alignment and the supporting metrics with several examples.
He also took on “The Myth of The One Version of the Truth,” and gave us the Law of Buytendijk: “The more a certain term is connected to the core of the business, the more definitions of it will be around.” He used a number of examples of ambiguous terms such as “revenue”, “flight”, and “student”, which are often defined differently by different folks within a given organization. Of course the solution was to drive out these different uses and add qualifiers to the terms. One could immediately grasp the importance of how these terms effect metrics and performance.
That afternoon, several friends and I took some time to take a very comfortable day-sailer out on San Diego Bay. Sailing in San Diego is always a big draw for me, and this was a glorious afternoon with just the right amount of sun and wind.
After doing a bit of writing in the morning, I wandered down to the vendor show. Having talked with some of the winners at the Best Practices luncheon, I wanted to follow up with Information Builders to see what else the company was bringing to market. The folks in the booth showed me a product called WebFOCUS Magnify which is based on a partnership with Google. Magnify combines the Google interface and search engine with Information Builders BI reporting, metadata, and heterogeneous data access capabilities to make high value data access just about as simple as I have ever seen.
While I was on the vendor floor, I decided to drop in on DataFlux. DataFlux offers a wide range of capabilities for managing data quality from profiling to monitoring. This last piece caught my attention. It’s one thing to find data quality problems and correct them, but it is another to monitor the arrival of new data for data problems. The DataFlux Data Monitoring server uses business rules developed in the process of profiling, which come from the initial discovery phase. The product then applies these rules to new data as it arrives. This is valuable, especially when you have no control over the original point of data collection.
That evening I attended Lance Miller’s night school session, “Enterprise Data Management—A New Perspective.” Miller gave us a good overview of EDM, but did it with a strategy that looked for problems that could potentially be solved through a sound data management program. A good software salesperson will look for a client’s pain, and show the client how his/her software will eliminate the pain. Good data management practitioners (and data warehousing/BI professionals) will do the same when it comes to selling an enterprise data management program.
Miller suggested problems to look for and architectures to help create a vision that would address the problems. Although several of my classmates seemed overwhelmed by the large volume of material covered, I thought that Miller had done a bang up job of providing us with a lot of valuable reference material.
Wayne Eckerson, Director of Research and Services for TDWI, led off the morning with a keynote titled, “Best Practices in Implementing Predictive Analytics.” Eckerson gave us a couple of other synonyms to level set his presentation—data mining, analytics, knowledge discovery. Data mining is how I got my start in BI after grad school, so I was pretty keen on what he had to say. Eckerson pointed out that these activities usually have a lot of moving parts, but have high business value. According to a recent survey conducted by TDWI, well over 75% of those organizations that responded either have, are developing, or are exploring this area. Eckerson went on to provide some insights as to what applications are most interesting—cross-sell/up sell, campaign management, and customer acquisition led the pack. He also reviewed what it takes to implement an effective predictive analytics program.
After the keynote, I dropped in on C. Lwanga Yonke’s session, “Leading Change: The People Dimension.” Yonke did a great job with the subject in presenting and leading us through several exercises. I usually like these sorts of classes because I find that BI problems are most often people problems. I have found that implementation of or enhancements in the BI environment need to be approached as changes that affect folks down, up, and across the food-chain. At the heart of many a failed DW program is a small IT group that has focused too much on control.
Revitalizing such a program is a challenge to the manger of such a group. It is also a challenge to rebuild credibility with the business side. Although much of Yonke’s material was designed to meet the needs of managers who needed to get their teams on-board, much of that material could be used to manage change as it affects those outside of his/her functional unit. The primary take-aways from this session were, there will be resistance and an open communications flow is essential to work through almost any change.
That evening I went to a peer networking session led by Larissa Moss. The subject, EDW (Enterprise Data Warehouse) versus PDW (Personal Data Warehouse) has been getting some attention in other conferences lately. However, the PDW concept is not new. Ken Orr wrote a white paper over 10 years ago that addressed the subject. The idea is that a PDW is created for a single or small group of sponsors, not the enterprise. This evening the group seemed to be split down the middle. Most EDW proponents came from IT, and most PDW proponents came from the business side. Both sides demonstrated passion. Although many of the problems surrounding the EDW, such as treating its development as a single, very long, and very big project, have been recognized as mistakes to avoid, I thought that the discussion was well worth having once again. It never hurts to keep each other focused on what is ultimately best for the business.
I wrapped up the week with Bonnie O’Neil and Lowell Fryman and their session titled, “Business Metadata: Putting Metadata to Work in the Business.” This duo was high energy and a bit slap-stick which was good, because this subject could use a little levity. However, as humorous was their style, they did offer some serious food for thought. They listed semantic barriers to understanding and put PowerPoints at the top of the list. These shows often subtly reflect the biases of the creator, imbed too many assumptions, and offer little attempt to create common understanding. O’Neil and Fryman gave us some basic definitions of business metadata, i.e. that which gives meaning to the business, business rules, goals, metrics, access lists, etc. They also pointed out some problems with poor meta data such as the meters/feet problem that caused a multi-million dollar Mars lander to miss its mark. Later they gave us some ideas about “harvesting” business definitions and how to promote meta data resources once we put a sound meta data collection and delivery infrastructure in place.
That afternoon as I flew back home, I started thinking about TDWI in Orlando in November. I read through a copy of the program and it showed a number of new or updated sessions. I also started thinking about kayaking with friends on some new Florida rivers afterwards after the main event. It promises to be an excellent conference. See you there.