I keep hinting – or saying outright — that I think dashboards need to be revolutionized. It’s probably time to spell that point out a little further.
The key issue, in my opinion, it that dashboards need to be much more personalizable than they are now. This isn’t just me talking. I’ve raised the subject with a lot of users recently, and am getting close to 100% agreement with my viewpoint.
One part of the problem is personalizing what to see, how to visualize it, and how all that’s arranged on the screen. No one product yet fully combines best-of-breed ideas from mainstream BI, specialized visualization tools, and flexible personalized web portals. But that’s not my biggest concern, as I think the BI industry is on a pretty good path in those respects.
Rather, the real issue is that dashboards don’t adequately reflect personal opinions as to what is important. Indeed, that lack is often portrayed as virtue, because supposedly top management can dictate through a few simple metrics what a whole company of subordinates will think and think about. (Balanced scorecard theology is a particularly silly form of this.) But actually that lack is a serious impediment to dashboard success, or indeed to a general analytic/numerate enterprise culture overall.
“One version of the truth” can be a gross oversimplification. We can laugh about how a company can have 5 or 10 or 13 different ways to define “revenue” or “profit.” But when that happens, it’s generally because multiple different ways of counting each have validity! For one thing, accounting for costs and reserves is an inexact science. Measuring overall value creation is less precise yet. For example, what’s the lifetime value of acquiring a customer? That’s almost impossible to know, yet estimating it is crucially important to almost any quantified sales or marketing decision.
Consequently, it is very reasonable to analyze based on alternate definitions side by side, each perhaps being proposed by its own champion. In particular, a middle manager should be allowed to keep multiple different versions of the same table, graph, or chart side by side, namely the one her boss thinks is right, the one she thinks is right, and the ones her subordinates favor. When all versions give essentially the same conclusions, all is good. When they differ – well, that’s a prime opportunity to sharpen the analysis and perhaps learn something new about the business.
In an important subcase of this, queries and visualizations need to be more flexibly specified (and specialized) than they are today. Here’s an example of what I mean. If there are general sales pipeline reports and graphs, district managers can easily be shown just the version for their district. But the need goes well beyond that. A sales manager can, himself, do a good job of creating a category of “important accounts to focus on” or “typical accounts (with special cases stripped out)” or “accounts that are likely to give major expansion opportunities down the road if only we get in the door.” But he has to be able to do it himself; it often can’t be handed down from above.
Since this is getting long, I’ll describe what I think needs to happen under the covers in a separate post.