David Raab Article Archive

Newspaper editors have a name for it–MEGO, for My Eyes Glaze Over. It describes topics that may be important, but are so abstract, arcane or otherwise removed from the daily lives of their readers that nobody will pay attention to them. Global warming, the Whitewater affair, and international finance are classic examples. Most people’s common sense approach is to ignore such matters until their immediate significance becomes apparent.

New computer industry buzzwords rank high on the list of MEGO topics. So while the term “Enterprise Information Portal” has been floating around for some time, most readers of this column probably have not given it much thought. Those who bothered to explore a bit probably decided that it was mostly a new label for existing end-user query and report distribution tools, brought together on a common starting screen. Since many vendors would gladly relabel their products to borrow some of the excitement generated by Internet “portals” like Yahoo!, no further explanation seemed necessary.

As it happens, this jaded view doesn’t quite do justice to “Enterprise Information Portal” (or “business portal”, as it is less formally known). The concept does have some substance, in that it extends beyond the traditional row-and-column databases of a conventional data warehouse to include unstructured data such as spreadsheets and text. It makes an implicit promise to somehow organize these unstructured sources so they can be searched efficiently. It further promises to incorporate external data sources, such as information on the World Wide Web. Like its Internet namesake, the business portal provides a single access point to all these varieties of information. And, at least in theory, it offers to correlate information from separate sources so it can be combined and compared in meaningful terms.

In short, the vision here is nothing less than one grand system that unifies all information everywhere. The IT community has a peculiar affinity for such panaceas, despite a long history of past failures. This alone is grounds for skepticism about the ability of business portals to come anywhere close to meeting their promises.

But there are more specific concerns as well. Since the purpose of a business portal is to integrate structured and non-structured data, it must at a minimum provide all the integration capabilities that a data warehouse applies to structured data alone. But as any data warehouse or marketing database veteran will attest, developing and maintaining the processes to consolidate data from disparate sources is by far the greatest technical challenge faced in warehouse projects. Developing analogous capabilities for unstructured data can only be more difficult still. In fact, it has so far been attempted primarily with techniques derived from “artificial intelligence”, a field that is itself notorious for failure to translate scattered initial successes into reliable, enterprise-scale solutions. This does not bode well.

Still more worrisome, the most scalable approach to organizing unstructured data involves automated processes that develop classification schemes based on similarities in document contents and usage. This is the exact opposite of data warehouse methodologies, which require careful analysis and planning by human beings. To expect that radically different methods will independently produce compatible classification schemes seems unreasonably optimistic.

In short, complete fulfillment of the enterprise portal vision requires that the vendors solve two of the most difficult problems in data management–consolidation of structured data and organization of unstructured data–with the additional constraint that the solutions must be mutually compatible. Since the vendors themselves must realize that success defined in these terms is highly unlikely, it is reasonable to ask whether they are actually pursuing more modest goals and whether the resulting products still offer something of value.

The answer to the first question is a clear yes; the answer to the second may be in the eyes of the beholder. Portal vendors with their roots in structured data, such as Information Advantage (www.infoadvan.com) and Sqribe Technologies (www.sqribe.com), focus on delivering personalized reports and analyses created with their traditional tools. Portal vendors with roots in unstructured data, such as Viador (www.viador.com) and Autonomy (www.autonomy.com), focus on creeating personalized lists of articles that are classified using their automated content analysis techniques. (Some vendors oriented to unstructured data rely on human editors to classify content and assign keywords. But this gets costly when volumes are large.)

The common thread that unites the two sets of portal vendors is the ability to create personalized Web pages that list information relevant to an individual. This of course requires a mechanism for individuals to specify their interests. Not surprisingly, the mechanism tends to be structured for structured-data portals and freer for unstructured-data products. In structured-data portals, users often are asked to select the specific reports, analyses or alerts they want to see. In unstructured data portals, users may define areas of interest in their own words or even have the system monitor their activities and adjust its selections automatically.

Both sets of vendors generally promise to incorporate both structured and unstructured data. After all, this is part of the fundamental portal concept. In practice, this usually involves some stretching to accommodate whichever data type is outside the vendor’s primary domain. For example, a structured-data system might create a database of standard keywords to provide access to unstructured text; users would then subscribe to a keyword the same way they subscribe to a report. An unstructured-data system might classify reports based on textual descriptions and the distribute the underlying tabular reports based on the results. At best, these approaches provide a loose integration between structured and unstructured data.

Still, personalized lists are the key capability that distinguishes portal products from earlier applications of the underlying technologies for data analysis or text classification. The true value of these personalized lists is difficult to assess. In many ways, they represent a less aggressive version of earlier “push” technologies that sent messages to individuals when relevant information appeared. Despite great initial excitement, “push” technology has not been a major market success. One might suspect that, ultimately, users prefer to decide for themselves when to turn their attention to different parts of their job, rather than having a system present data for all possible tasks simultaneously. If this is true, the success of business portals may ultimately have less to do with the actual presentation of the lists than with the quality of the data analysis and classification tools that lie underneath.

Eyes glazed yet? Readers who have made it this far are probably wondering what business portals really have to do with relationship marketing. The true answer is, not a heck of a lot. But some vendors have suggested that portals might somehow replace traditional marketing systems, so it’s worth understanding enough about portals to deal with this topic should it arise. If you find yourself in this situation, bear in mind two key points:

$ the real value of a portal system lies in the underlying applications. So if you’re offered a portal system that includes a campaign manager or customer analysis functions, you need to assess the quality of those tools themselves. What’s important about a door is what’s behind it.

$ portal systems are mostly built for data access, not operational execution. So while a portal might be useful for a marketing manager or analyst, it would not usually replace a campaign manager or customer contact solution. Although a portal could theoretically provide access to such systems, they would most likely be distinct products, unintegrated with other portal functions. Any claims to the contrary should be examined very closely.

David M. Raab is a Principal at Raab Associates Inc., a consultancy specializing in marketing technology and analytics. He can be reached at draab@raabassociates.com.