I am at the KMWorld conference in Santa Clara.

This presentation, "Enterprise Information Architecture: A Framework for Intranet Success", is by Louis Rosenfeld. It is actually a panel, and has Louis; Debora Seys, the information architect for the @hp Employee Portal, Hewlett-Packard; Joseph Busch of Taxonomy Strategies, and board member for the Dublin Core Metadata Initiative; and Mike Crandall, the technology manager for the U.S. Library Program of the Bill & Melisaa Gates Foundation.

I am a big fan of Lou's book, Information Architecture for the World Wide Web, second edition. I had to choose between this presentation and another one I wanted to attend called "Enabling Knowledge Discovery: Taxonomy Development", by Jayne Dutra, the Knowledge Management Process Owner for NASA Jet Propulsion Laboratory.

Lou just asked how many people in the room were information architects, and most of the people raised their hands. Snikies!

Lou is up: What's Wrong with Intranets. Users' perspective. User can't find what they need. Silos that look like the org chart, which are used business areas that own content. Org charts change every 2-5 months. We are demanding that they think the way the company is organized.

Lou's slideshow is online here.

This is also a problem for the people who run the site. They are stuck in politics, cultural issues, and technologies. They often have only a few months to do a whole IA.

Users' problems + organizations' problems = enterprise information architecture (EIA).

EIA framework: seven issues. 1) EIA governance (how the work and staff are structured). Ideal: a separate business unit. Usually gets stuck in IT or Marketing, which each has its own baggage of how it approaches these issues.

Needs to balance strategic and tactical issues. Strategic: model on Board of Directors. Tactical: start with staff who "do stuff".

2) EIA services (how work gets done in an enterprise environment). Services group to the enterprise. Chinese restaurant menu is a good model for IA services, not monolithic service model. Set of services that solve problems for internal clients. Assess departmental IA needs (content acquisition, content authoring, indexing, metadata, search index, controlled vocabulary, etc.).

Lou is suggesting an entrepreneurial model, what he calls intrepenurial. Can do business planning, can do projections based on demand. Offering basic and premium services, and projecting trends for these.

3) EIA staffing (who handles strategic and tactical efforts). IA is basically users matched to content in a business context. IA staff can come from a variety of disciplines. Ideal team would have IA generalists focused on big picture, some IA generalists focused on more local business unit issues, and also IA specialists who really know things like thesaurus design, search engine indexes, etc.

4) EIA funding model (how it gets paid for). Look at other centralized cost centers in the enterprise, like HR and IT. Can those models work for IA. Hybrid, diverse model, between central and business unit funding. Seed capital (from central fund) plus (from business units) flat tax plus variable income from special projects and from premium services.

5) EIA marketing and communications (how it gets sold inside the company). If IA doesn't have good buzz, you can use other terms like knowledge architecture and user experience. Use what name that matches what internal client is looking for, that is within the scope of IA.

Break things into pieces that are manageable and digestible. It is difficult to take on a complete EIA. Start with low-hanging fruit. Look for influential users and projects with strategic value in the business context.

One way to illustrate: show EIA that is working already. Employee directory could be a good model.

6) EIA design / timing. Says we are trying to turn the Titanic with a rudder consisting of our foot. Scope is 3-5 years.It takes time. We want it to happen faster and more consciously, so there is more organizational learning.

Faster is really less slow at this point. But, it is happening as part of the natural evolution of the enterprise.

Taking a modular, phased approach. The stuff that happens soon, that stuff that happens later, long term, etc. Some have to do with top down IA, and some with bottom up. Some guerilla IA, e.g., blogs, k-logs, etc.

Use early successes as models. Do things early that can be done quickly.

7) EIA workflow questions (how does content authoring and publishing work). How does content get from someone's head on to the enterprise intranet.

Content workflow goes from decentralized to centralized. Authoring is decentralized, but categorizing content is more centralized (on an enterprise intranet). Different business units handle decentralized vs centralized differently, and they each need to be accommodated. Need to help them dedicate resources to decentralized processes, and help you dedicate resources towards centralized IA processes, like creating controlled vocabulary.

Debora is up. She is talking about her team which is in the solutions team in the @hp portal organization, which is dealing with three different programs: the portal, intranet management, and content management. Each has its own business model.

@hp portal's business model is based on integration. When people want something to appear in the portal, they integrate with the portal. They go through process of complying with requirements, usability review, reliability standards (stable site with good sponsorship and authority), and also implement the portal's look and feel and design guidelines. In particular, they have a home page template for organizational department home pages.

They have an entire account management organization managing integrating organizations into the portal.

The intranet management has a model based on zones. Segmentation model based on types of content on the intranet. Standards and requirements are based on segmentation model. Critical and authoritative content used by executives, or very critical to the running of the company. Very rigorous zones. All the way down to the ski club, which is not required to comply to many standards.

The content management program has a business model based on engagement with contents. Standard project template for implementing CM for internal customers. Engage like a partner to create a CM solution, based on Documentum. Has metadata and controlled vocabularies embedded in it.

Help implement standard IA metadata things in CM implementation.

All of this was developed at HP over a number of years. Developed to get the most done with HP's loosely autonomous departments.

Joseph is up: taking some contrary points of view. Most organizations fall into the non-innovative approach to content organization. They may be good at data management, but they are struggling with content management (or unstructured data).

He defines IA as bringing together disparate content into a common interface, regardless of data structure.

He is usually working in organizations where there is a long way to go to get to the model Lou was talking about. He thinks they are good models, but thinks a lot of us have to live with less sophisticated organizations.

EIA Goal: a single point of access to all enterprise content. Address issue of stove pipes or silos. He likes to use the term content architect rather than information architect. He is trying to connect it to data, which has value that can be quantitized and related to the business case.

We are trying to throw-over the silos in a meta architecture, i.e., with metadata. We leave things in place where they exist, because we can't control them. But, we put tools in place that create a single point of access (some times called a portal). Most important thing: this single point of access.

Mike is up: raining on Lou's parade. Ideals don't apply in real world, in most organizations. Changing the way organizations are now is a difficult thing to do. Lou's examples are great as ideal models, though they probably don't work in the real world.

Mike was at Boeing, and didn't get very far. Former Boeing group at new company (recently) and had the ideal plan, the right pieces, the right people, buy-in by senior management. But, budget changed and the whole IA process got rolled-back. It didn't last.

Talking about Rob Cross' Social Network Analysis and energizing interactions. A large part of the success of IA projects is due to working social networks. A lot of it is about who you get energy out of to make something happen. Without these, even good IA work won't get very far.

Those interactions outside of IA structure are important, and can be leveraged to help success. When he was at Microsoft, a lot of people were competing to come up with solutions for IA type issues. He worked to transform the competition to a cooperation, and took the best ideas and resources and got them working together. This was successful.

Resources are the thing that most affects IA projects. Good connections get good resources, to keep the IA process going. If not, IA process gets stopped, because it is long-term and takes time to show major value.

Being tactical about how this approached in critical. It does impact all the areas that Lou mentioned. Mike hadn't thought about it this until hearing Rob speak, but thinks it is a good way to look at it.

Q & A

Q about data architecture initiatives getting connected up to IA initiatives. Example, Federal EA initiative. Joseph answers: noting the connection is a first step.

With the Federal initiative, it is top down. The real question is how is it going to work. There are standards, but it is not clear how it is going to be driven with applications.

It serves us well to engage professionals know about this (metadata) already, in terms of ERP and data warehouses. IA is part of a long tradition of information management, and it shouldn't be invented from scratch.

Q from a content manager at Allstate in audience: trying to get unstructured data management problem to be appreciated as addressing a context of data management problems. Data people didn't get it. Had to tag along with portal team to finally get some recognition, since the portal dealt with content management.

In the context of the portal, the content management is starting to get recognition as part of the data architecture effort.

Debora comments: at HP, recognition of collaboration, content management, and portal frameworks as having overlaps. Unstructured data gets more structured in CM systems, and that has begun to be recognized by data architecture teams.

She also is talking about the business case. You need to tell a story that people kind-of get. Inside HP, the story was library with books on the floor and needing to clean it up. But now, need to come up with a new metaphor to energize it. Translate strategy into a story, and have story as basis for ROI that people can get at an emotional level.

Q from a content management specialist in the audience: talking about TeamSite and Verity being used. Now working on common metadata and taxonomy strategy. Senior management doesn't see value in it. Even brought in Meta consultant, but not convinced.

Mike is talking about Boeing. How IT shop and its traditional models of data were in charge. Now, five years later, there is a healthy interaction between this and people who come from the content world. They are learning from each other: content people learning from data modelers, and vice versa. This interweaving is very important. Example: semantic web, sign of the conversations happening.

From the audience: recommendation of Zachman Institute ebook on enterprise architecture. Talking about structured vs unstructured data terms, reference management is the metadata issue.

Joseph recommends book: The Content Management Bible, by Bob Boiko. Lou recommends Ann Rockley's book Managing Enterprise Content. Also, Tony Byrne's CMSwatch website.

I ask a Q of Debora about portal vs intranet at HP. A: Problem with boundary issues. Now that there is the intranet management process with segmentation and zoning model, we are starting to get a sense of how to balance issues, like search problem (quality of content accessible by search engine). Registration application to register their websites.

HP - Compact merger pushed some of these issue forward. They had to merge both companies' portals and intranets. Did some clean-up as a result of that.

Recognition of portal from top down, used as a stick for getting people to follow standards.