Tuesday, August 01, 2006
non-executive dashboards
Some follow-up to my posting last week on productive dashboards.
Interaction designers need to consider the different needs of traditional "executive" dashboards, which provide a big picture of activity throughout an enterprise, and is typically packed with minute data, and employee-centric dashboards, which reflect information relating to the activities of an individual employee or a small team they belong to.
To see what is happening in enterprise dashboards, one should consult a fantastic resource, The Dashboard Spy. Nearly all the examples posted are executive level views -- generally amalgamating and summarizing unit level data that used to be handled by Excel spreadsheets. Many these dashboards look like Excel spreadsheets. (Happily, some such as SAS/GRAPH have improved on jarring appearance of Excel's default graphics.) One would expect the mountains of data summarized, and capable of being sorted, drilled-through, and otherwise manipulated to be useful in analytical decision making such as business intelligence. If one's role is to make decisions for others, such by-the-numbers visualizations can be a powerful aid. It is still possible to drown in such data, be given so many options to manipulate it that one's never such if one's seen all the views one needs to see to make a sound decision. But executives are paid to make such decisions, and often they worry they aren't able to see things from all angles and link disparate variables. We'll defer to the brilliance of the executive to figure out what he or she needs, and not worry about the UI being too complex. On balance, access to many variables with many "degrees of freedom" to manipulate these variables is useful for executives.
Now let us consider line staff. Their job is more concerned with doing things, rather than thinking about how other entities should be doing things. This distinction is sometimes murky with the rise of self-management. As employees are often responsible for making their own decisions, it may be tempting to think they need a mini-executive dashboard. But the more the UI forces them to think about what they should be doing, the more distracted they can be from doing it.
To illustrate the employee's dilemma, let's look at a demo dashboard created by Visual I/O. Visual I/O is an exciting player in dashboards, creating visual user interfaces with impressive graphical representation and interactivity. The demo on their website concerns baseball; it is a playful illustration of interaction concepts and presentation methods they apply to more mundane subjects. They have designed a dashboard to manipulate variables to assess whether a pitcher should be replaced during a game. It reflects an extreme case of the fetish some baseball fans have for statistics!
Let's play along with the scenario, and imagine someone actually using the dashboard. While I can imagine a manager toying with the dashboard the next day at his desk, trying to figure out patterns for how to best rotate pitchers, I have trouble imagining him doing these tasks in the dugout as the game is being played. At that point, he isn't following the game, he's absorbed in the fantastically data-rich user interface. Now, let's stretch our scenario a bit and imagine that the pitcher also uses the same dashboard, which he had loaded on a PDA in his back pocket. Between pitches, he would toy with the dashboard, trying to figure out if he should ask to be relieved. The likely deterioration to his pitching concentration would present the answer, regardless of what the dashboard data suggested.
What pitchers and others involved in performing tasks need are simple heuristics to make decisions, not mountains of data. Dashboard design could learn from the field of heuristics. A useful volume on this topic, recommended by Don Norman, is Gigerenzer and Todd (eds) Simple heuristics that make us smart (Oxford, 1999.)
Interaction designers need to consider the different needs of traditional "executive" dashboards, which provide a big picture of activity throughout an enterprise, and is typically packed with minute data, and employee-centric dashboards, which reflect information relating to the activities of an individual employee or a small team they belong to.
To see what is happening in enterprise dashboards, one should consult a fantastic resource, The Dashboard Spy. Nearly all the examples posted are executive level views -- generally amalgamating and summarizing unit level data that used to be handled by Excel spreadsheets. Many these dashboards look like Excel spreadsheets. (Happily, some such as SAS/GRAPH have improved on jarring appearance of Excel's default graphics.) One would expect the mountains of data summarized, and capable of being sorted, drilled-through, and otherwise manipulated to be useful in analytical decision making such as business intelligence. If one's role is to make decisions for others, such by-the-numbers visualizations can be a powerful aid. It is still possible to drown in such data, be given so many options to manipulate it that one's never such if one's seen all the views one needs to see to make a sound decision. But executives are paid to make such decisions, and often they worry they aren't able to see things from all angles and link disparate variables. We'll defer to the brilliance of the executive to figure out what he or she needs, and not worry about the UI being too complex. On balance, access to many variables with many "degrees of freedom" to manipulate these variables is useful for executives.
Now let us consider line staff. Their job is more concerned with doing things, rather than thinking about how other entities should be doing things. This distinction is sometimes murky with the rise of self-management. As employees are often responsible for making their own decisions, it may be tempting to think they need a mini-executive dashboard. But the more the UI forces them to think about what they should be doing, the more distracted they can be from doing it.
To illustrate the employee's dilemma, let's look at a demo dashboard created by Visual I/O. Visual I/O is an exciting player in dashboards, creating visual user interfaces with impressive graphical representation and interactivity. The demo on their website concerns baseball; it is a playful illustration of interaction concepts and presentation methods they apply to more mundane subjects. They have designed a dashboard to manipulate variables to assess whether a pitcher should be replaced during a game. It reflects an extreme case of the fetish some baseball fans have for statistics!
Let's play along with the scenario, and imagine someone actually using the dashboard. While I can imagine a manager toying with the dashboard the next day at his desk, trying to figure out patterns for how to best rotate pitchers, I have trouble imagining him doing these tasks in the dugout as the game is being played. At that point, he isn't following the game, he's absorbed in the fantastically data-rich user interface. Now, let's stretch our scenario a bit and imagine that the pitcher also uses the same dashboard, which he had loaded on a PDA in his back pocket. Between pitches, he would toy with the dashboard, trying to figure out if he should ask to be relieved. The likely deterioration to his pitching concentration would present the answer, regardless of what the dashboard data suggested.
What pitchers and others involved in performing tasks need are simple heuristics to make decisions, not mountains of data. Dashboard design could learn from the field of heuristics. A useful volume on this topic, recommended by Don Norman, is Gigerenzer and Todd (eds) Simple heuristics that make us smart (Oxford, 1999.)
