Links & More

	Print version

Related Links
	Portals – The Information, Functionality and Services I Need at My Fingertips
	The Posture of Portals
	Short profile of Karen Holtzblatt
	Contextual Design at SAP (Holtzblatt & Beyer)
	Review of Contextual Design (Beyer & Holtzblatt)
Background Links
	Incontext Website
	mySAP Enterprice Portal

Creating New Work Paradigms for the Enterprise Portal

By Karen Holtzblatt, InContext Enterprises, www.incent.com – 05/21/2001

Disclaimer: Please note that this edition was written in 2001. Therefore, statements in the articles, particularly those regarding SAP's products, product strategy, branding strategy, and organizational structure, may no longer be valid.

The Web started as a way to share information and documents between people at a distance. As information was "put out there" for general use by professional groups, both within and across companies, we took the first step towards using this new technology to support enterprise workers.

Today's Enterprise Portals can support greater numbers of people in the workplace than ever before. Several factors came together to create this opportunity:

Most enterprise workers ignored overly complex desktop tools - but the Web's ease-of-use pulled in managers, sales people, and anyone else looking for information.
'Click and go' interaction with the Web is simple. At its most complex it requires no more than 'fill-in-the-blank' and click.
For the vast majority of workers, work focuses on information use, communication and coordination. None of these activities require anything beyond point-and-click complexity.
The Web overcomes technological hurdles that have plagued companies for years - enabling database integration, storage-independence, and centralized system support, all across distributed organizations.

As Web usage has become increasingly ubiquitous, people are recognizing its potential to support work practice within the Enterprise Portal itself. Information is only one, admittedly large, part of the needs for enterprise work support. Today the Web is inviting movement from an information-centered purpose to an enterprise work-support purpose. But we cannot take advantage of this opportunity until we overcome two major challenges:

How can we improve access to the massive amounts of information available over the Web?
What kind of interaction paradigm will allow work information to be integrated into applications that support daily work activities?

In this article, we'll discuss a new approach to organizing information-'neighborhoods of knowledge'-that provides direct access to information, with easy and natural exploration. We'll integrate this with ideas for organizing 'push' information that comes to the user, and Web applications that support work tasks. The result expands the role of an enterprise portal into that of a full work support tool.

Getting Useful Information is Harder than Ever

The problems of finding information are age-old. People have invented numerous ways to access information, through libraries, databases and individual folder structures. Even so, the challenge of finding information has yet to be solved, especially in today's situation.

People are already lost in a sea of information. Some is pushed to them directly; some must be searched for in contained "pools" such as CD-ROMs, databases and buildings like libraries. The Web allows everybody to put up their own pages within and between companies. Email pushes supposedly relevant information and communications to us directly.

Today we have access to unlimited information. But this unlimited access exponentially increases the difficulty of finding useful information.

Current Search Techniques are still Ineffective

The solutions we have today, search engines and categorization, too often do not work:

Search engines provide, in exchange for a short phrase, a disembodied list of links, related only because they each use the search phrase somewhere in their pages. Our user research confirms what everybody knows: that most people only look at the first few hits, that there is no guarantee that the links are really relevant to the user's actual intent, and that the address titles do not give a good idea of what the user will find if they take the time to look at the listing.
Categories allow people a more targeted way to 'window shop' for information that will meet their intent. But people don't know the rules of the organizational structure and whether what they think the category means is in fact what the category means. People have to look through many categories and build a model of how it was put together to use it effectively. And they still don't know what a document link will really bring up.
'Good Samaritan' pages collect links by the intents or needs of a user community, assembled by someone who shares that intent. They are better than a category structure because the person maintaining the page is a known member of the community. He or she replaces the search engine's disembodied list with a more coherent, targeted list-if you trust the good Samaritan. But how do you find the good Samaritan page in the first place?

Each of these solutions leads to what we call a 'leggy' search experience. People blindly click to go one place and the next moving from one category to the next or one page of links to the next. One page leads to another as the users drill through lists of links hunting for the content they desire. Searches lead to search results which lead to more search results or more pages of links. But our research shows that people are only comfortable going two clicks away from their starting point because they don't want to get lost and they don't want to wait for pages to load.

Figure 1: A 'leggy' site such as this makes people travel blindly to find what they need. However, people don't want to go more than two clicks from their starting point because they can get lost and load times are slow. (Click image for larger view)

Blind Clicking is like Driving Behind a Truck

Blind clicking, with only a title or single line of text for direction, might be acceptable if pages came up in nanoseconds and you could flip through them as easily as a magazine. But pages are too slow to load and they fill up the whole screen. The experience is more like driving down the road behind a truck. You can't see where you're going. You know you're going forward but you don't know what's coming up and you don't know if you'll eventually get where you want to go. When we are driving behind a truck, we look for a way to go around it so we can see ahead. Every Web page is a truck blocking the user's view ahead and preventing them from seeing how to go forward to the information they want.

Natural Scanning Uses a Personal Focus

At InContext, we studied how people look for information in the real world. We observed people in a variety of venues: reading newspapers, looking at magazines and newspapers, doing research for papers or to answer questions, teaching, touring museums, and watching documentaries. We found that people use a personal focus to scan through a 'neighborhood of knowledge.' They start with a small pool of collected information (a neighborhood) that hangs together according to some known intent: a newspaper, a defined database, a pile of papers they collected on a topic, the shelves of one section of the library, an area in a museum. Once in the neighborhood they scan the topography-they see the content laid out in space and scan for relevance and interest, stopping at headers, abstracts, bullets, pictures, known names, tables and other quick indicators of content.

The power of the personal focus is clear if we can just get people to the right neighborhood of knowledge they can quickly "flip through" it using their personal focus. This is our challenge: How do we provide the right neighborhoods of knowledge for the users of any portal? How can we use their personal focus to help them scan and "look beyond the truck" to see where they need to go?

Figure 2: When we scan through pages to see what we want to read, the words that reflect our personal focus 'jump out' at us, even from large blocks of text. (Click image for larger view)

Structure Information by Work Intent

Help People Get to a Neighborhood of Knowledge

Our research suggests the most powerful solution to finding information is to bring people to a contained neighborhood of knowledge that they can scan with their personal focus. But this requires understanding the fundamental intent of the users. If we understand what workers want and need, we can structure access points to information. These access points organize neighborhoods of knowledge relevant to users' work intents. An access point is not a portal. It is a window into an organized set of data available at the top of a site. Like the display window of a store, it reveals what's inside so users can decide if they are interested. Providing multiple access points allows users to scan across them and find the one that matches their current need. This is very different from using topical categories in a library or taxonomy. Once you know which store has the kind of clothes you want, you can go in and browse the racks to find exactly what you need.

Access point design requires that we understand why people are coming to a site and what they want to do there clearly enough to define the core access points that should be revealed at the top. It also assumes we can reveal what is in a neighborhood of knowledge a user is about to enter - whether this is the neighborhood that contains the desired information. Current interfaces, whether lists of links or categories, usually give only titles, one-line descriptions, or topics. None of these provide adequate information to "see ahead" of the truck.

Flatten the Site Structure

For a portal, the challenge is to flatten the structure, eliminate legginess, and raise the content to the top of the site. Chunking the information by intent sharpens the focus of a set of information for the user. Then the challenge is to reinvent topographical scanning - help users move quickly over large amounts of information, following their personal focus to determine where to go. Access points organized by intents can both describe what is beneath the access point and raise content all the way to the top. We can now provide roll-over information that is deeper than a title, to reveal where the access point will take the user and what a document is really about without having to click to see. This flattens the site structure and reduces legginess without losing the organizational structure of the content.

A portal organization that works will raise the necessary content up to the user in a form that supports natural scanning. Such a design lays out information on one page. Users can scan and choose the part of the site that matches their intent. A page may provide access to multiple neighborhoods or may be the neighborhood itself - in which case it may provide both views into the content and annotated access to related neighborhoods.

Rollovers are effective for bringing content to the top level, letting users quickly scan for the part of the site

Figure 3: Rollovers are effective for bringing content to the top level, letting users quickly scan for the part of the site that matches their intent.

Structure the Overall Site to Reflect Work Intent

The notion of access points assumes that the portal is constructed as a coherent environment, taking into consideration how it will be used. Within large organizations, almost invariably different teams and departments design pages that organize and represent their work without thinking about how the site hangs together as a whole. Standardizing the navigation framework and the branding won't keep the site from being leggy and unstructured from the viewpoint of people coming to the site with a particular intent. Designing a single site is different from the Web itself. A site can be thought of as a building or shopping center in a city - it is planned for usage. The Web is more like a city that grows new buildings and roads - yet even a city has ordinances and city planners who look at the overall utility of the city for transportation and living.

An enterprise portal is contained enough to allow for this planning. Navigation standards are necessary - but planning the fundamental structure is also necessary. An intent-driven structure is not a structure driven by pages and links, nor is it a simple reflection of the organizational structure - each department or team with its own portal. A person trying to find information doesn't want to walk the organizational structure, trying to guess where something needed might be. An enterprise portal is there for the use of the people within the enterprise - so it is possible to identify the intents of the workers engaged in different kinds of work. Structuring sites by intent will yield quicker access to information because the site structure will match the user's entering intent.

Page Layout is the New Challenge for Consistent UI Design

Watching people scan for information, we realized that the structure of the content page itself can aid scanning and quick information access. There is only a limited set of content types for knowledge workers: product descriptions, requirements documents, corporate news, white papers, etc. But typically, the layout of these pages is not consistent across the company. People using this information will go from page to page to page, stopping on each page to reinterpret the layout before they can 'see' the information they want. Each different type of information carries its own tone and natural structure, yet different teams re-invent the layout. The result is that people can't find what they're looking for when they're scanning. Just as newspapers have a standard layout paradigm that lets readers scan for news quickly, we need to define standard layouts for different types of information on the Web so people can recognize the page type and be able to scan its structure for the information they need.

Understanding the fundamental intents and practices of people using, finding and scanning information has led us to design new solutions and interaction paradigms appropriate for optimizing information use within enterprise portals.

Organize Pushed Information like Your Desktop

The information the enterprise sends to you creates an entirely different problem. Going out to the Web or Intranet causes overwhelm because it is hard to find what you want. But email, which is supposed to be relevant to you, is also overwhelming. Many people, especially executives, get hundreds of emails a day. Some of the information is important, but most is extraneous mail from bloated distribution lists. Folders, which should organize messages so they can be found again, hide their contents instead. Information is sent over and over because it's so easy to lose in the email organization. People lose critical email in the flood of general messages, in the folders they create as they try to stay organized, and in the clutter on their desktop.

People's physical desktops suggest a more coherent model. Invariably the space directly in front of a person is kept empty. Immediately beside that is the most current work, which they pull into that empty space to do the work. Half an arm's length away is the 'what I'm going to work on this week' pile, and one and half arm's lengths away is the 'stuff I ought to read pile.' Behind the user is likely to be a shelf of reference material.

The designs InContext is working on with SAP make these distinctions salient. Alerts come to you with a look that shows you they require immediate action. Notifications relevant to your work group themselves by topic and intent. General information to help you keep up to date is separated from action items generated by the system or in meetings. News about the company or outside world is separated again. Information appears in different places in the page or portal, according to content.

Role Identification Based on Work Intent Makes it Possible to Push Useful Information

To do this the system must know who you are and what you are doing. That means it knows something about your job and your key work activities and intents. This is the power of role definitions when applied to portal design. A system that knows what you are likely to be interested in and what you're working on can guess what kind of information is relevant for you. It can push information relevant to you as defined by your job and your team membership when you are hired, in addition to the information you subscribe to explicitly. This is like creating an active access point - a combination of automatic smart push and your own subscription to and collection of information turns your portal into the optimum access point for everyday work intents.

Identifying and tagging information appropriately requires understanding users' roles, tasks and intents. Job titles alone don't tell us what a person does. Instead, think of people as wearing a set of hats when they do their jobs. A hat is a role they play to get part of their job done - a set of responsibilities which will achieve a job intent. Roles imply both information and action. If we can represent the roles that support an organization and identify the ones each user plays, we can send them the right information without clutter. And if we can present it in a way that is as natural as desktop piles, people won't get lost in the sea of information. As shown in Figure 4, mySAP is already beginning to offer content according to users' roles.

Example where first level tabs correspond to roles, not job titles. The second level tabs are the tools, information and places the user needs to do the work of that role

Figure 4: In this SAP design, the first level tabs correspond to roles, not job titles. The second level tabs are the tools, information and places the user needs to do the work of that role.

Integrate Information with Work Practice

We have said that if we want to get to information we should organize sites and individual portals by the user's intent. But their intent is driven by their work. Information is a dominant dimension of all work practice; there's no work that doesn't involve creating, consuming or sharing information. But the vast majority of work isn't about creating information - it's about communicating, sharing, discussing and passing information. Information is shared continuously in every enterprise. Formal and informal meetings and collaboration keep the enterprise going, supporting decision-making, creativity, clarification, and contexts for assigning tasks. In meetings and other work contexts using information dominates the activity - which suggests that if we provide information on the Web we could support the rest of the work and integrate the whole work context.

The Web Can Support the Majority of Enterprise Work

The Web's ability to traverse space makes it the natural media for supporting distributed, collaborating people and teams. More and more enterprise workers are distributed - they need real-time coordination support and asynchronous access to information. Real teams, ad hoc teams, work groups, and sales people with their customers can work together over the Web because their work is not the kind of creating that requires specialized tools (unlike coding or building product). Even creators like developers spend much of their time coordinating, sharing and passing around what they have created for comment, discussion and review. As a result, the Web can be the platform to integrate real work activities and the information needed to do them:

Workflow coordination, task assignment, signatures, forms, and even paper reviews are about handing off documents for review. Any kind of tracking and passing along of information, which is natural for an enterprise system, can be done more easily on the Web.
Getting status and other information for decisions is almost the most important thing that drives companies. This kind of information can be shared more easily on the Web.
Because the Web is a good context for integrating multiple forms of information into a single place, it's a natural organizing medium. It allows us to bring together our emails, calendars, and documents from multiple locations.
Task items driven from a system instead of a face-to-face meeting, could be designed into a single place.

To support these kinds of work activities, think in terms of smaller applications integrated with the information that users need, and displayed in an organization that matches the work. At InContext, we are designing such small applications and making them available in portals tied to the roles that people play and the intents they need to achieve to do their jobs. Some examples:

Meeting places with the tools to create agendas and run the meeting, along with online access the information participants need.
On-line team rooms that collect needed information and access to people in one place.
Workflow applications, both formal and informal, that track where things are so users can see what's going on, do their part, and push the work along.
Task assignment tools to help project leaders distribute tasks.
Targeted, relevant information for daily status, so all information about work is coordinated in one place.
Self-organization tools, such as personal, company, vendor and business calendars.

Move Beyond the Information Interaction Paradigm

Bringing these small applications together challenges our Web interaction paradigms tailored for information page display. The navigation bar in a column on the left is standard for information pages, but does it apply to applications? We are not navigating from page to page - we are moving from context to context, stopping at a context to get a piece of work done. A small application needs more function than page display - it needs the screen real-estate to make the application work. Portal standards that insist on left-side navigation bars and tabs inhibit the evolution of the Web into a work support platform. Personal portals tailored to the roles we play will provide not only information, but applications based on our job activities. This move to the Web as a work support tool should change how we design the framework of the portal, the page of the portal, and the navigation through the portal.

As technology evolves, enterprise workers will be given information and tools to do their jobs through their portal much as you are given a computer and software today. People will be able to pull together their own unique work sites, not by consciously articulating what they want, but by seeing a valuable component out on the Web and pulling it into the right slot in their portal. The industry is moving in this direction; enterprise portal designers should be ready to support a full work support environment.

Design from an Understanding of Work Practice

What does it mean to design an enterprise portal today? We believe portal designers must have a deep understanding of the work practice of the people who will use the environment. Whether we are talking about how to access, traverse, and scan information, or what kind of application can be available and integrated with information on this platform, success comes from understanding what people are doing and why they are doing it.

Portal design is different from application design because it is not the design of a single task-it is the design of a whole job environment. It requires understanding a person's whole job. People will always need to tune their own environment, much as they decorate and move around the things in their houses. But they need not build their houses, their working context, from scratch.

If we understand the basic roles and intents of people in enterprises we can come up with useful layouts for typical jobs and load them with the appropriate information and applications. So today more than ever there is an even greater need for customer-centered design: customer-centered enterprise portal design driven by a deep understanding of how people work.

top

Archive - Edition 3: Portals

Leading Article

What is a Portal?

What's in a Portal: MiniApps, Generic MiniApps

User-Centered Portal Design

Graphic Design and Branding

"Real" Portal Projects