[ Home ] [ Up ] [ News ] [ FAQs ] [ Downloads ] [ Products ] [ Beginners ] [ Services ] [ Resume ]
[ Feedback ]
Hello and Welcome to my web site.  Thank you for coming. Best Regards, Peter Nolan.

Quotable Quotes from Metagroup

Metagroup publicly publish 'Metabits' which is basically a series of short comments.  When I was searching around for interesting comments I found it a little hard to find what I thought where the most interesting 'quotable quotes'.  I figured other people would have the same problems so I have, with Metagroups permission, copied some of what I think are the best 'quotable quotes' here.

Obviously the 'Metabits' are the opinion of Metagroup.  Where I have felt it is appropriate I have added my own comments in italics.

Data Consolidation: Consuming the Elephant

As IT organizations embrace the benefits of an enterprise data warehouse approach, the struggle to accomplish the audacious goal of consolidating all corporate data can be overwhelming. META Group research on analytically mature companies suggests that this large problem is actually a series of small projects, with smaller and more tightly scoped efforts making acceptance easier. Many never even announced a data mart consolidation policy. They simply attacked it one project/subject at a time. The key factor in getting and keeping funding was the IT organization's ability to consistently produce deliverables (new subject areas) on a timely basis. This kept success stories rolling in, built credibility for IT and the consolidation effort as a whole, and maintained momentum. Charles Garry

This has been in line with my experience.  Most consultants will recommend that you approach building an enterprise wide data warehouse one project at a time. However, you do need to build the initial infrastructure such that later components can be 'plugged' into the EDW.  As far as I am aware there is no truth to the rumor that you can just build a suite of star schema data marts and somehow they will magically integrate. Take for example the customer dimension.  There is a lot of work to make sure that the code that builds the customer dimension can take feeds from many source systems yet still produce just one customer record.

What You Don't Measure You Can't Manage

Process orientation drives intensive number gathering in enterprises. However, it is important to find the right key performance indicators. For example, there is little value in counting the number of calls answered by a call center. Instead, the number of calls with first-call resolution should be counted. Normally, numbers in categories like time, cost, and quality are collected and summarized. To achieve real improvements, it is necessary to maintain links between the gathered numbers to the original processes. Ruediger Spies

Lord Kelvin said something similar. I must look up that quote.  I have been a strong advocate of 'closed loop' business measurements and 'closed loop' marketing analysis for a long time.  I am also a strong supporter of the idea that motivating change within an organisation needs to be augmented or associated with a set of measures. The most critical parts of organisations already know the importance of accurate measures, sales and finance.  It has always amazed me that other areas of large organisations can get away with relatively obscure or no measures are all.

Best Practices of the Analytically Mature

Through 2004/05, most organizations will revisit their current analytic infrastructure to improve overall analytic agility and reduce IT related costs. IT organizations will lead this movement by spurning data-mart-only strategies, opting instead for more normalized enterprise data warehouses. Successful organizations will find that achieving sustained innovation driven by improved analytics will require passion on the part of IT to facilitate such change, and a leap of faith on management's part to visualize the long-term and profoundly positive impact of a data mart consolidation effort on the organization. Charles Garry

I expect that this comment will generate quite some discussion on dwlist.  I guess it depends on what Charles meant by 'more normalized'.  There is little doubt in my mind that organisations will be looking to be able to support faster changing measurement models, or want to measure things that have not been measured previously.  One of the questions that experienced consultants look at is 'What if a new measure is required, and it needs to be calculated backwards in time, say for the last 13 months?'  This is driving us towards storing more historical data in some kind of archive such that new measures can be developed backwards in time.  Many customers I have worked with have chosen not to implement this kind of archive early in the data warehousing process because the costs were high and the benefit unknown.  I see more and more people starting to realise and understand the benefit of such an archive. Of course, you can only go back and develop new measures to the point in time that you started archiving historical data.

I have also been a pretty strong critic of the idea of developing separate data marts and trying to consolidate them at a later point in time.  I have been a strong advocate that the data modeling in a large data warehouse should be done by one person, or at most two people, and there should be a permanent employee to take the work forward.

Analytical CRM in Europe: Differences All Over

The adoption rate of analytical CRM varies across European countries. META Group research shows that Scandinavian enterprises are most likely to have analytical CRM systems in use (38%). 33% of companies in Italy, UK, and Spain use these systems for customer data analysis, while in Germany just 28% have analytical CRM systems installed. However, 59% of German companies indicated they plan to implement analytical CRM this year. Ruediger Spies

The interesting thing about this comment is that the vast majority of organisations surveyed do not have analytical CRM systems in place. Having implemented my first such effort at this 11 years ago and seeing the very large profit improvement from such simple efforts I do wonder why people are not doing more about analytical CRM.  To do an initial cut today is quite inexpensive.  If you are not afraid of re-work or evolution of the system you don't have to buy a million euro of hardware and another half million euro of software.  

Data Integration Market: More Doers, Fewer Dollars

Due to the persistent need of organizations to integrate data from an increasing variety of data sources, we project the data integration (DI) market (e.g., extract-transform-load [ETL] and enterprise information integration [EII]) to grow at a pace outstripping many other technologies. However, because of downward price pressures introduced by DBMS vendors (most notably Oracle and Microsoft) offering low-cost DI solutions, DI market revenue gains (10%-15%) will fall short of actual adoption-rate growth (25%-30%) through 2003. Doug Laney

One of the reasons I have published the data warehouse generator on this site is that there is no doubt in my mind that there is no money to be made from it. The ETL tool vendors have now developed their products to such an extent that the entry level functionality for an ETL tool is quite high.  Microsoft is going to dramatically improve DTS and it is going to remain very, very cheap.  This is going to cause price pressure on the ETL vendors.  We have already seen the first generation of ETL tools go the way of the dinosaur.  Prism, Carleton, ETI, fine tools that they were, have gone or been consumed by other vendors.  Customers have voted that there is no place for high priced ETL tools, even when they were overwhelmingly cost justified. The data integration problem has not gone away, if anything it is getting bigger and more difficult than ever, it's just that customers have refused to pay a lot of money for software to solve that problem.

Customer Info Latency Approaches Escape Velocity

Customer information infrastructure at most organizations has been designed to manage low-velocity customer data. However, this customer information infrastructure cannot cope with the wide spectrum of dynamic information now demanded across the enterprise. During 2001/02, the trend has been clear that data warehouses are becoming more operational as business-critical components of enterprise solutions - 24x365 and increasingly less batch-like. By 2003/04, market-leading enterprises must be able to react intelligently and instantly to changing customer information via "real-time customer data integration." Near term, choices for such infrastructure range from semi-batch operational data stores (IBM CIIS, Oracle OCO, Siebel CIF) to near-real-time customer data integration engines (Apama, Journee, Siperian). Aaron Zornes

The change in requirement for timeliness of integrated data is one of the biggest changes taking place in the data warehousing industry that I am aware of. It is very, very hard to take data from multiple operational systems and integrate it in any sensible and valuable way very quickly.  The best way to do this is to make sure the operational system is aware of the key to be used during the integration and passes that key for every change that occurs.  But large customers with many systems will find it very difficult to change them so that the integration process is passed data a short period of time after the change takes effect.

Integrated Analytics Require "Deep Analytics" or "Data Primacy"

Although most BI vendors are headed down the "integrated analytics" path, they have the disadvantage of owning neither the "deep analytics," such as SAS and SPSS data mining and predictive models, nor the data primacy/ownership that the mega ERP and CRM package vendors enjoy (Oracle, PeopleSoft, SAP, Siebel). By 2004/05, the dominant vendors for integrated analytics will include: Business Objects, Cognos, Hummingbird, Oracle, PeopleSoft, SAP, SAS, and Siebel - with each of these vendors dominating certain vertical industries. Niche vendors will thrive by targeting specific industries (e.g., SpotFire in pharmaceutical and oil and gas, Manugistics in warehouse management). Aaron Zornes

"Integrated Analytics" for Breakthroughs in Corporate Decision-Making Capabilities

Increased competitiveness and other business decision-making pressures related to economic stress are mandating that business intelligence (BI) deliver information to the people who need it now. The marketing "noise level" will increase markedly in 2002/03, due to vendors jumping on the marketing bandwagon for "integrated analytics," as each BI vendor attacks the others' technical features (e.g., Business Objects, Cognos, SAS). Through 2003, "integrated analytics" will increasingly amalgamate BI components into a single, end-to-end analytical framework that can serve all the diverse user requirements across the enterprise to solve specific decision management problem areas. Aaron Zornes

Of late I am pretty convinced that analytical applications, like the ones from Cognos, Business Objects and Informatica will be very attractive to the medium sized enterprise on a relatively limited budget for their informational systems. Some of these analytical apps have quite comprehensive measures built into them and can source data from quite a few of the ERPs.  However, for large companies the ERPs (eg. SAP) are heavily customised and there are usually still many home grown systems.  There are also the cases where different ERPs have been purchased.  So for large companies I am at somewhat of a loss to understand how a suite of pre-built analytical applications will be of great use given that the bulk of the data warehouse build is in the ETL process.  Further, I would expect that a significant amount of customisation will be required to the analytical applications to include different, customer specific, measures. 

BI and Mega Applications: When Worlds Collide

Enterprises have invested millions of dollars/euros in ERP, CRM, and SCM solutions, yet remain frustrated by their inability to receive aggregated, timely information from their new operational systems. The mega-application vendors are increasingly competing on the basis of embedded analytics (e.g., Oracle's "virtual close," PeopleSoft's Enterprise Performance Management, SAP's Business Information Warehouse, Siebel's eBusiness Data Warehouse). These vendors also show a determination to compete aggressively with their former BI partners and optimize the revenue capture associated with these formerly add-on capabilities. During 2003/04, the intersection of "analytics" and "application packages" will be of great interest to all software vendors, systems integrators, and investors in these two dynamic, churning markets. Aaron Zornes

Over the last 5 years I have been surprised time and time again that companies who bought ERPs believed that they will provide a single integrated source of information for decision support.  Every time I heard a customer say that I just couldn't believe my ears.  These customers had just not done even a nominal amount of research on what is possible. When I explained to customers that they were not going to get such a decision support system I rarely got a good showing.  These vendors are going to tout their 'integrated analytics' for years to come, and I hope they improve, but I fail to see how an ERP vendor is going to come up with an 'integrated analytics' suite that will readily accept data from competing ERPs. Further, as the ERP vendors start moving into this space I fail to see how the analytical application vendors Cognos, Business Objects, Informatica et al are going to be able to continue to co-operate with the ERP vendors in developing new releases of analytical applications for new releases of the ERP.  It will be the easiest thing in the world for the ERP to change data sourcing rules or field names under the covers so that the analytical application vendors product no long functions as required.  I shall wait and watch this market with some interest.  

Why Companies Need a Knowledge Worker Infrastructure

Not designing for collaborative interactions as part of business requirement gathering results in subsequent IT system implementations lacking contextual integration of a user's work activities, forcing users to bounce back and forth between collaborative and non-collaborative systems. Worse, even within a given set of collaborative tools, users have an inconsistent experience as they traverse multiple environments. Portals are a critical first step in establishing a consistent (unified) framework for knowledge worker infrastructure (KWI), and portals will increasingly become role-based and part of business process integration efforts at the user workspace level. Mike Gotta

Let's face it. In the 80s I was working for IBM selling the concept that applications should be designed off the base of a single enterprise model and all applications should talk to each other effectively. Of course, we were selling the idea that the single enterprise model should be on an IBM mainframe and preferably DB2. We lost that argument. 

Vendors of ERP systems came along with a promise, purchase integrated applications and all your problems will be over.  Well, only if you purchase ALL your applications from the one vendor, and even then many would argue that all your problems of data integration are over. Large companies buy their applications from many vendors and these vendors are not going to collaborate any time soon. So their systems are not going to collaborate any time soon. So the scene is set, I sincerely doubt that we are going to see 'collaborative' systems for transaction processing any time soon. Having lost the argument about building integrated operational systems in the first place I now find myself in the position of championing the building of integrated informational systems in the second place. Quite amusing.  

Users have got what they asked for and bought, an inconsistent user experience as they traverse from system to system.  I don't see that going away.  Nowadays, if users want a single integrated source of information to measure the company by the only place they are going to get it is from the data warehouse. Funny how some things work out.     

Analytical CRM Spending: 25% CAGR

To improve sales and rationalize major deployments, application vendors (e.g., Siebel, SAP, E.piphany) have begun heavily marketing analytical capabilities within their own offerings. Global 2000 organizations are receptive, because analytical CRM budgets are increasing. We believe analytical CRM spending will see a 25% compound annual growth rate, reaching $5 billion by YE06. A strong analytics program will enable business to differentiate prudent customer-focused initiatives from specious projects that overspend and distract the organization from its core mission. Kurt Schlegel

I only hope that people spend money on useful analytical CRM projects focused at generating more profit for their company and not just more profit for the vendor.

Obstacles to Effective CRM Analytics

Analytical CRM deployments will face numerous challenges stemming from a lack of expertise, data modeling complexity, and proprietary control of information within various business segments. In particular, most vertical industries will struggle to combine analytics for offline and online channels. During the next two to three years, insufficient analytical spending will result in CRM failures, as some organizations will be unable to identify customer opportunities or empirically communicate the value of CRM programs. Kurt Schlegel

Lack of appropriate expertise has been a major problem in the data warehousing space since I first started. For a while there anyone who had any experience with any collection of data called themselves a 'data warehousing consultant'. Many of the 'disasters' I have been called in to kill or fix have been the result of inexperienced people.  I expect the situation will be the same for analytical CRM applications.  Personally, I have no problem with inexperienced people creating disasters and wasting money.  As a group, they demonstrate the value of a consultant who actually knows what they are doing!!! 

From Murky "Data Puddles" to Clear-Running "Rivers of Information"

Business intelligence (BI) is the iterative analysis of business data via an organized, coherent method, typically using online analytical processing (OLAP) capabilities - e.g., calculations/aggregations across multiple dimensions. Increasingly, the requirement is to provide a single consistent interface to access corporate data - whether in multidimensional OLAP cubes or in multiterabyte (TB) relational databases. Hence the flurry of OLAP integration with relational database systems (e.g., IBM UDB with Essbase, Microsoft with Analytic Services, Oracle 9i with Express). The objective is to seamlessly combine multi-TB access with the speed of multidimensional OLAP cubes for reporting, query, drill-back to source operational data, and statistical analysis. By 2004/05, the "integrated analytics" test for larger enterprises will be the degree to which heavy-lifting analytics such as data mining are incorporated. Aaron Zornes

I have been bemused by the whole argument about ROLAP vs. OLAP for a long time. I have sold and implemented many data warehouses that contained both ROLAP and OLAP tools. Why? Because there are some things that OLAP tools are really good at.  Like Essbase for financial reporting using Excel as a front end. I worked on a deal with MicroStrategy where we stood up at the client and told them they would be crazy to try and do financial reporting through Excel using MicroStrategy just as they would be crazy to try and perform ad-hoc drill anywhere analysis of millions of records using Essbase and Excel.  It nearly worked, the customer finished up buying Business Objects for the more ad-hoc analysis.  There is space in this world for OLAP cubes and ROLAP tools sitting on top of star schemas. The two are not mutually exclusive and we will see more of this in large companies.

BI/Data Integration/Analytic Tool Integration Delivers Business Value

During 2002/03, business intelligence, data integration, and analytic tools will become more information-focused, concentrating on providing business value rather than technological prowess. By 2003/04, comprehensive (a.k.a. integrated) information delivery and analysis architectures will become available, delivering both unstructured and structured information through enterprise portals. Through 2006, further integration, including business application vendors embedding tools such as analytic engines into their products, will enable optimization of customer-related business processes and data quality, thereby reducing “information sprawl” and yielding substantial business ROI. Louis Boyle & Doug Lynn

This is a very interesting prediction.  There is no doubt this is the direction the vendors are moving in, can they do it so quickly? Let's wait and see.

[ Home ] [ Up ] [ News ] [ FAQs ] [ Downloads ] [ Products ] [ Beginners ] [ Services ] [ Resume ]
[ Feedback ]

Send mail to peter@peternolan.net with questions or comments about this web site.
Copyright © 2002 Peter Nolan
Last modified: October 28, 2002