Pattern from Barbara Mirel

GOAL:  Determine a competitive product mix and persuade stakeholders to implement it.

QUERY-AND-INTERPRET activity

 Subgoal:  Establish a baseline for sales and growth


Figure 1. Partial set of patterns for Establish a Product Performance Baseline.

Establishing a baseline for sales and growth is one of many QUERY-AND-INTERPRET subgoals in which users are engaged when determining the best product mix for their store. Users derive this baseline from the sales and growth of the 3-5 top performing products in a category.

The first step in establishing a baseline is to get an overview ? a 30,000 Foot view. Users look at all products’ sales and growth by manufacturer, brand and item for each outlet of sales, region, and market. This 30,000 Foot View pattern embodies a number of patterns (#1-7 in figure 1), each involving its own group of patterns. See Figure 1.

The patterns named in Figure 1 only go to a certain level. In reality, each of the lowest level patterns in the diagram embodies even more finely grained patterns, working down the hierarchy to single actions with individual interface objects and to the under-the-hood relationships among objects. I stopped at the level above because I see it as a critical Janus-faced  point. It is a this level that the patterns turn at once to users’ large-scale real world goals and motives and to  their direct manipulations at the program interface.

From Figure 1, the sample pattern that I’ll describe is Information Flow for Precision. This pattern comes into play when users need information to flow from the interface to tell them the meanings of the graphics and precise values relevant to setting a baseline for product performance. As Figure 2 shows, graphics used for the 30,000 Foot View present multiple charts and a data sheet, giving users a workspace for discovering  standards relevant to a successful productg mix. Missing from immediate sight in this workspace, however, are ample information aids ? precise values of data columns and bars and labels for all items. Users need this precise , complete information for valid decisions and persuasive effect.. A description of Information Flow for Precision follows.
Figure 2. Graphics for 30,000 Foot View.

INFORMATION FLOW FOR PRECISIONS PATTERN

Information Flow for Precision plays a role in many subgoals of QUERY-AND-INTERPRET ? it is a core activity in interactive visual analysis. One challenge in designing for Information Flow for Precision is that the informational content, presentation (structure and appearance), and visibility that users expect and need vary by work context and data table within the subgoal of establishing a performance baseline and across diverse QUERY-AND-INTERPRET activities. The design of Information Flow for Precision, therefore, inherently embodies a tension between customized and one-best-system (generic) approaches. The following description reflects our group’s understanding of the needs, problem, and solution of this pattern so far.

Context:  The graphics for a 30,000 Foot View immediately and perceptually show a good deal of vital information about high and low performance through size, height, color and position. Users, however, need more precise information as well because their interpretations and conclusions will affect the success or failure of their businesses. Users need precise and accurate values for data points and aggregations and labels for data items. The purposes served by this information and the times during inquiry that users access it vary and overlap. Users need this information for such immediate inquiry problems as follow:

Users’ problems relevant to this pattern are straightforward. Addressing them in software and interface design is more complicated since content, presentation, and visibility depend on purpose, context, and data.

Problems and Contending Forces

Relevant Content -- Users expect relevant information on an as-needed basis with little effort or interruption to their ongoing visual analysis. Yet without a lot of built in intelligence, the program cannot presuppose all cases and is apt to either over- or under-supply users with information. Users also expect the information to be “chunked” according to their integrated way of thinking about their problem and analysis. Making discrete facts available and letting users pick, choose and combine them as they like distracts them from their train of thought, diverting attention from their task to the tool.

Contending Forces: Combined content relevant for one situation may be irrelevant for another. Providing different ‘flows” for different cases lead to too many  types of flows for users to keep in mind. Providing discrete facts for create-your-own flows undermines problem-solving efficiency.

Presented Right ?.  At a point of need, users expect to see almost simultaneously visual and textual renditions of the data. They want to read it easily (legible), extract what they need immediately from its format, and dock it where it will not interfere with the graphical data. If too much information is crowded in a small space, users can neither access information easily nor get it to display legibly in printed reports or notes. If due to size, form, or lack of mobility, the presentation of values and labels blocks the visual display, users cannot perform core actions such as selecting data, rotating, it, and so on.
Contending Forces: A small size that does not block graphics may be illegible, and it may inhibit providing enough information. Printed presentations are a different “genre” from presentations relevant to as-you-go analysis. Information flows for data of interest, when docked, lose their data context.

Visible As Needed ? At times, users want to display precise values or labels continuously; other times, they want to “turn them off” in order to devote more screen real estate to graphics. Because having this information available as-needed is core to analysis, users do not expect to devote much if any attention to this present-or-hidden choice. Any design that shifts attention to the tool (e.g. many keystrokes, window positioning and rearrangement) annoys users.
Contending Forces: Continuous visibility reduces the space devoted to graphics and, at a point if it is context-sensitive, leads to perceptual clutter.

Solutions

Provide multiple flows, some that automatically appear, others that must be accessed. Provide context sensitive precise information that appears and disappears when cursored-over.  It remains visible as long as the cursor is over the data so that it can be printed with a screen shot. It does not display when users use the cursor and left mouse button to select. At that point more reduced information on values/coordinates appears off to the side in order not to interfere with users’ field of vision for selection. Provide as well a Stats Table window accessible from the toolbar that docks in the bottom left hand corner of the screen. This table gives key information on the total value, the percent of the whole comprised by the selected data, the number of cases selected and deleted.  If the default information is not right for users, they can select in Preferences from a range of options, the combinations and relationships they want to view.

Examples

Providing the Stats table improved users’ ability to visually establish a baseline because, for instance, they could sweep the top 5 brands in a bar chart, press the toolbar button for the Stat Table, and immediately see on it (already calculated) that these brands comprise 85% of all market sales. The availability of this information increased the efficiency of this activity. They left it docked for the rest of the visual analysis. In addition, seeing labels/values of items appear off to the side while sweeping over bars to select them resulted in users’ reported confidence in the accuracy of their analyses rising from “Somewhat Confident” to “Very Confident.”

Other Possible Solutions

Ideally, as we gain insights from more actual use cases about what information users need for various types of problems and analysis, we will more elegantly design content, presentation, and visibility with minimal elements able to address multiple siltations. At the program architecture level, the program can differentiate among types of analysis ? e.g. simple descriptive analysis answering what questions; analytical statistics answering why questions; over time analyses) ? and automatically structure and populate information flows in the Stat Table to coincide with the specified type of analysis.

References

Information Flow for Precision is part of many other pattern sets across subgoals in solving the problem of finding and convincing others of a competitive product mix. Within  the one subgoal of establishing a performance baseline it is part of other patterns such as Sweep-and-Select, Monitoring, Notes/Reports, What Data Tables, Fields, and Ranges (in Data-to-Graphic Mapping).

 


An example pattern prepared for the Usability Pattern Language Workshop at Interact '99