In our travels to meetings & conferences, something we hear a lot is “That sounds great! Can you send me your dashboard?”
Now… I don’t want to minimize the awesomeness of a well-structured and accurate dashboard. They’re easy to understand, they’re attention-grabbing, and they’re concise. But they’re not really why you’re digging into your data. I want to talk about what dashboards are good for, and what that means for the kind of data problems you’re probably trying to solve.
What Dashboards are Good At
Dashboards reflecting cross-system data within your organization do two important things. They call attention to outliers and they support research into what’s really going on.
Think about the metaphor of the dashboard: a car’s instrument panel. My car has 36 sources of information on its dashboard. Only 4 of them are gauges that provide me with real-time data on my car’s operation. I really only look at those to check my fuel or when there’s a police officer in my rear view mirror. The other ⅞ths of the data are status and warning lights.
A dashboard is good for calling your attention to something anomalous. When one of those warning indicators lights up, you want to know about it so you can react to it. Check out this tiny graph:
Imagine this is your graph. You’d have one of two reactions. You might say “What in the world happened at 2:15 this morning?” Or you might say “You can see exactly when homework is due.” Either way, it behaved as a warning indicator: WARNING: Problem at 2:15 or Situation Normal. You know which one it is because it’s your business. In the right hands, dashboard widgets like this can help diagnose abnormal situations and get everything back on track.
You Don’t Need to See My Dashboard
Just as you can’t drive your car by looking at my dashboard, you can’t get insight into your organization using a dashboard that was developed for mine. In most organizations, it isn’t even the same person who wants to (or is allowed to) see all of the anomalies, so the concept of “one dashboard” is of pretty limited use.
What you really want is to put a well-tuned warning light in front of the right person so they can correct a problem or make use of an opportunity. That requires knowledge of the questions being asked and how to interpret the answers. It requires understanding of what situations are likely to arise and what interventions will be effective. And it won’t hold still; it will change over time. You’ll discard things that don’t work and you’ll tweak the things that show promise.
In other words, asking to see my dashboard won’t help you. It will tell you if I have a good designer on staff, or maybe if I’m color blind, but to really make use of a dashboard, you have to make your data accessible, make sense of it, and put it in front of people who can do something with it (sound familiar?). I’d be very surprised if my dashboard worked for you, but I’d believe you could use one of your own.
In a future post, we’ll go into how to build a dashboard that’s customized for your situation without doing the tuning the hard way.