Building Bridges Platforms Matter

FRAMINGHAM (03/15/2000) - Let's clarify one thing before we begin. User departments do not hate the IS department. They do however find us to be generally unresponsive to their needs and concerns. This may be due to our own arrogance or their lack of understanding. Nevertheless, be it fact or perception, the result is sometimes a relationship of mere tolerance.

My 26 years as an IS professional have given me some insight into the reasons for such mutual animosity. Ready or not, here they are: the IS department's lack of emphasis on being service oriented and the IS professional's lack of emphasis on people skills.

Does this mean that we are not service oriented or that we totally lack the skills necessary to relate to our fellow beings? No. But we do not emphasize their necessity in our everyday work environment. This responsibility falls on IS management.

Studies have shown that the dominant personality type in our profession can be classified as introverted. Funk & Wagnalls Standard Dictionary defines an introverted person as: "One who is sober, reserved and withdrawn." We as IS professionals tend to be more concerned with Microsoft Corp.'s latest release of the Windows operating system than with the business problems and concerns of our users.

IS management must be held accountable for defining and enforcing our true purpose within the business community: to help our users achieve their business goals through the use of information technology. Management must realize and believe that the responsibility for open and effective communication with user groups is a two-way street that should be constructed and maintained by the IS department. We must be bridge builders who understand that communication among departments is absolutely essential to our success.

We need to add elements to our definition of the successful, competent IS professional. In our hiring process, IS management should evaluate prospects not only on their technical expertise but also on their social IQ--their level of sophistication with respect to social interaction. We need people who have technical know-how mixed with diplomatic and salesmanship skills. We need people who are willing and able to interface with user departments not only as technological advisers but as business partners as well.

I place a lot of the responsibility for the success of the IS-user relationship on the IS department. The reason for this is quite simple. The IS department knows what's going on with respect to information technology, and the user community, for the most part, does not. We must assume the role of mentors toward the user community. True mentors do not belittle or alienate their protgs but have an innate understanding of the learning curve they must negotiate and the anxieties they encounter. We must provide the users with the facts necessary to make prudent information systems decisions without undue criticism or unfair comparisons to ourselves. We must be personable, approachable and responsive. The user should be treated as a valued customer on whom our livelihood depends. These attitudes will work wonders in closing the rift between IS and its user communities.

Ruffin Veal III is the IS associate manager for Ramsey County in St. Paul, Minn.


Throughout its history as a unit of corporate organization, information technology has functioned like a two-stroke engine: First we do production, then decision support. In plain English, first we make systems work--payroll, billing, shipping--and then we figure out what happened.

Until today, this model has served us well.

But it's time to move beyond this serial model to one where these activities run in parallel. "Business Intelligence" systems examine what happened after the fact: Who were the top five salespeople, which products turned over fastest and so on. They predict the past, always lagging behind the curve. That mind-set will not help us drive new business opportunities, create new markets or uncover competitive advantages.

What we need instead are "Intelligent Business" systems that understand and optimize the present--and predict the future--so that we can leverage its implications. And we need these systems today, because today's successful competitors no longer follow the "transactions first, decision support second" model.

Let's look at how profiling (the collection of information about a user as an exemplar of a category) and personalization (the development of behavior-based, unique information about individuals) work together on a successful e-commerce site.

During your first online visit to your favorite bookseller, you provided personal information. The vendor stored that data, compared it to purchased demographic information and categorized you according to credit-worthiness, likely preferences, price sensitivity and so on. You described yourself as an erudite reader with a fondness for 19th century European histories, poetry and Mozart. This profile drives specific offers to you the next time you log on.

Classic decision support--multiple file lookups, joins, decision trees and retrieval of products and prices--is based on your profile. No transactions have yet taken place in this interaction. Sophisticated model? No, it's already old hat.

It's the next level that requires data of a new kind--clickstream data describing what you actually do on the site. It's a step beyond point-of-sale information because it's personalized; in fact, it's behavioral analysis. More sophisticated tools, with larger volumes of data and complex weightings, are employed. In our example, it turns out that your behavior is not as sophisticated as your self-image--you've been buying cheap adventure novels, cartoon collections and heavy metal music. (Let's be kind and assume your teenage son has been logging on as you.) If the system is well designed, it will give more weight to your behavior than to your profile, leading to more sales.

The effort to achieve such sophistication is driving radical change in tools and architectures. Data mining capabilities are moving into databases to support sophisticated, active decision processing that leverages the information infrastructures built in the data warehousing wave. Portals facilitate the parallel streams of decision support with transactions in a way that rejects the two-stroke model. If you are not investigating these technologies, you're falling behind. Make this year the turning point, and abandon the old model before you are as obsolete as it is.

Merv Adrian is vice president of Giga Information Group, a research and consulting firm in Santa Clara, Calif.

THE FOLLY OF PLATFORM INDEPENDENCE By William f. Zachmann It's been fashionable in recent years to deny the importance of fundamental IT hardware and software platform choices. According to the popular wisdom of the day, one ought not be too concerned about underlying platform choices but to concentrate instead on "solutions."

This, like so much other so-called popular wisdom, is far more popular than it is wise. Senior IT executives who uncritically accept claims that platform choices do not really matter are sure to pay significantly more to get significantly less value in systems for their organizations. The truth is that platform choices are among the most critical choices CIOs make.

The crux of the argument that basic hardware and software platform choices do not really matter lies with three seemingly plausible yet thoroughly misleading claims. The first claim asserts that senior IT management should concentrate on solutions and leave the rest to technicians. The second claim holds that various forms of middleware make underlying hardware and software platforms irrelevant. The third claim states that the cost of basic hardware and software platforms is such a small proportion of total IT spending that it really doesn't matter which platforms are used.

It is true that an organization really needs solutions in the form of information systems that enable it to compete in a rapidly changing environment. It's not true, however, that solutions are--or can be--developed wholly independent of the particular hardware and software contexts that enable them.

All too often the devil is in the details. The details of what can or cannot be done with a systems project and--even more important--how easily and how speedily it can be done, will depend very much on the specific hardware and software platforms used. Repeating the solutions mantra to the exclusion of careful consideration of platform choices is a sure way to fall prey to those devils.

There is, to be sure, a growing range for so-called middleware options. They range from basic standards like HTML web interfaces, through common frameworks like object-oriented development and relational databases, to more sophisticated would-be inter-enterprise standards based upon XML and the various proposed business object library standards. There are certainly constructive opportunities to take advantage of all of these.

To imagine that one can do so at the enterprise level, to the exclusion of basic platform choices is folly. First of all, none truly offers, or will offer anytime soon, a comprehensive framework for information systems for the enterprise as a whole. Many promise to do so, but none actually delivers on the promise. Even the best contain a considerable dose of snake oil and promised "miracle cures" for your IT ills that they have no real ability to deliver.

When you get down to it, the practical implementation of systems built on some sort of middleware still depends upon the real capabilities of the underlying hardware and software platforms on which they rest. Systems built entirely on supposedly higher-level, allegedly platform-independent middleware are systems with their feet planted firmly in the clouds.

The third claim--that basic hardware and platform costs do not matter--is as self-serving on the part of those who promote it aggressively as it is wrong.

It is analogous to the equally ridiculous claim that, for practical purposes, communications bandwidth might as well be considered infinite and free.

Hardware and software platforms are not free. There are substantially different costs associated with different platform choices, not only direct costs but much greater indirect costs as well.

It is no accident that the most aggressive advocates of the "platforms do not matter" argument are also the IT vendors that are trying to peddle the most expensive hardware and software platforms on the market today: IBM, Sun, Oracle and other vendors currently under the guns of Microsoft Windows NT/2000's advance into the enterprise on Intel servers.

Stripped of the abstractions that disguise it, what the "platforms do not matter" crowd is really saying is: Do not use Microsoft Windows NT/2000 on Intel-based servers. Instead, keep paying us traditional premium prices for old-line IBM mainframe and semi-proprietary Unix-based midrange servers.

For any enterprise that really wants to get the best value and best results in industrial-strength information systems today, that is not very good advice.

William F. Zachmann is president of Canopus Research, an IT market research and analysis firm in Duxbury, Mass.

Do you have an opinion you would like to express? Let Senior Editor Megan Santosus know at

Join the newsletter!


Sign up to gain exclusive access to email subscriptions, event invitations, competitions, giveaways, and much more.

Membership is free, and your security and privacy remain protected. View our privacy policy before signing up.

Error: Please check your email address.

More about CanopusGiga Information GroupIBM AustraliaIntelMicrosoftOracle

Show Comments