There’s been a decent amount of conversation surrounding yesterday’s post from Anil Dash, “All Dashboards Should Be Feeds”. It struck a nerve with me because it’s something I’ve been contemplating over the past year. To a certain extent I agree with Anil, but let’s be honest, his example of a substitute feed isn’t perfect.
The real conclusion that Anil should have drawn isn’t as controversial: we desperately need data analysts to provide context. To dismiss dashboards is ridiculous though. Dashboards are powerful for those data analysts who are capable of using them to produce derivative content that’s contextually relevant.
The problem is that in a world of big data (information overload), derving context is challenging and can go horribly wrong. Here are just a few ways dashboards can fail:
- They don’t always provide context - In other words, a chart is only valuable if the person who developed it can use it to surface relevant information. Too often we create pretty charts that provide little insight. In other words, dashboards are reliant on people who can use them effectively.
- Sexy can be distracting - The rise of data visualization has created a large market of people who can make pretty charts without any insight. It creates noise in an increasingly complex world, the exact opposite of its goal (which is distilling complex information to make it easily digestible).
- Information overload - A close friend of mine who runs a successful internet startup showed me the dashboard he looks at to see how things are performing. He monitors countless data points in an effort to monitor how changes are impacting their bottom line. The more data points he tracks, the more likely he is to miss something. But there are plenty of strengths in dashboards. Let’s say that you are collecting a billion data points per day. A dashboard would be useful for getting a high-level overview at what’s happening and help you find what you may be missing. You can then further boil down the information into bite-size chunks that people can use to quickly digest what’s important and what isn’t.
In other words, the dashboard can support analysts who produce the summary feeds that Anil Dash describes. Yet just like dashboards, summary feeds are only as good as developers and analysts make them. When the summaries aren’t comprehensive, what will people turn to? Dashboards.
I have no doubt that summary feeds will become an increasingly popular tool for distilling information. In fact, dashboards and feeds will become increasingly connected. Yet to demote dashboards to anything below a “supporting information interface” is ridiculous. Each interface has its own role in a world of information overload.