They shop the same, but they actually BUY things at the end of their shopping... 
I totally agree with you about follow up being necessary and even agree with the generalization that a form lead captured on a research site is weeks if not months out and Cars/ATC and the dealer site are ready now. The stimulus for the contact is usually a specific piece of inventory.
The data does matter. Here is the problem Rob and all other dealers have to solve for in the internet marketing era BEFORE cutting budget lines. As the medium grows and matures we are quickly discovering that our ROI calculations are often dreadfully wrong. Look at CPC on a banner ad for example. There was a time not too long ago that we thought you could measure effectiveness by cost per click alone. We've since learned that may be the worst way to determine real value. There are just a ton of other KPI's to consider.
This response is getting long already, but I was trying to get at the need to figure out what sources are influencing the purchase while a prospect bounces their way to the bottom of the funnel, NOT just the last trackable actions they took. I was tracked as a website sale on my last purchase, but I sourced the car I wanted through a classified site and never would have gone to the dealer's site if I hadn't been introduced to them on the classified site. I did exactly what I wrote as an example.
I guess what I'm saying is really just the ZMOT concept. The influencing sources during the search and research are equally as important as the lead submission source.
Now how does that fit into cost per SRP difference between ATC and Cars? It may not, but the tracking survey I was talking about may show you that cost per SRP shouldn't be weighted very heavily in your formula for the real value of that source.