[parasubvert]

I feel you. Time tracking analytics is God awfully complicated and mostly impossible to get right unless you have a very disciplined culture. I've railed against such poor systems in my younger years.

But sometimes there are ways to make it work. At Pivotal Labs for example, you are allocated to a pair daily and then to a project or product over a few weeks. There are no interruptions or meetings beyond townhalls and daily stand ups... Unless you and your pair occasionally want a ping pong break. Productivity and resourcing becomes a bit more reasonable in this culture as its related to stories/requirements in Pivotal Tracker being completed relative to the number of pairs rather than tying it to individual worth. It's like an incredibly kind and collaborative culture that also gets good data.

Stepping back to a general management perspective, To me the most effective data is if you keep it relatively simple and do controlled experiments that are systemically tied to an external market result. A lot of companies in the logistics (and now software services) space get good at this.

[jaawn]

That is a much better approach. It incorporates the humility you need to have with this type of data. That system works because it does not assume and expect accuracy, it works around the problem in a different way.

So, I guess my original sentiment is more like: "data-driven" is not a good thing on its own, because you need more than just a bunch of data, you need analytics.

It may seem more semantic than anything else, but I think there are some real differences in how people (especially managers) perceive "data-driven" vs. some other term like "analytics-driven". "Data is not magic" should be a catch phrase spread far and wide among the non-technical business world.