Is There Too Many Variables to Draw Signals?

In the past, we kind of ramped spend too early in my opinion. We didn't have our framework for experimentation set up properly. And so now I'm just trying to spend more intentionally and know what kind of signals to look for. It's really interesting when you have high spends going on and you market one feature and you market another feature and one feature of seven demos and either one drove one, you spent the same amount of money. So that's sort of the process, but we'll be happy to answer a follow up because I just kind of stayed high level there. Bob is going to represent an ABM question as well.