100% agree with the theory of what you're saying, but in practice I've found that none of these new event-based analytics tools provide much value until you add custom attributes to the events. That is where you get the important slicing and dicing and segmentation.
Unfortunately, Heap's approach only captures the events automatically but can't capture the business-specific attributes that need to go on those events. Ultimately that will always require some human being to think about their business-specific problems.
1. Empirically, we've found that our customers rarely fall back on custom attributes. The vast majority of queries run in Heap (>75%) operate on automatically-captured events. To me, this suggests we've either: 1) cut out a significant portion of implementation work, or 2) enabled analysis that was previously blocked by implementation work. If either (or both) are true, it's a win, and suggests that automatic event-tracking produces salient data out-of-the-box.
2. That said, some metrics are important and do require manual instrumentation. Coincidentally enough, we're about to launch a feature that solves the problem you mention and requires no extra implementation work. We're excited about it. Want to try it out on newrelic.com?
Unfortunately, Heap's approach only captures the events automatically but can't capture the business-specific attributes that need to go on those events. Ultimately that will always require some human being to think about their business-specific problems.