Hi Jeff, new member here. Seems like a great community...
Have long Lurked around, but never signed-up, so yesterday I finally did.
The information and learning here is excellent.
On to your post... Having been in the Lead Gen. and Handling space for years, I thing the general question is a valid one. Getting consumed on nickels and dimes will only distract you from respecting a valued part of the Sales process at most Dealers. That said though, most dealers today have a multitude of Lead Sources and from an efficiency point-of-view need to "some how" manage this multitude so they are aware of the well-performing and poor-performing sources.
Funnily enough, most Lead retailers will talk about a 'blend' of Quality vs. Quantity, and I think it is the 'quantity' part that serves as the MOST annoying to Dealers, both from a $ and Time perspective. But the formula otherwise, just distributing Quality, would not be tenable for the Lead Gen's sites out there. They need the revenue of reselling the poor-quality a few times over to aid profitability because the cost of only Quality leads would seem offensive to many Dealers.
The other approach would be to simply look at the total-spend on a Purchased Leads and compare that to revenue generated. If that number is good, then why worry. Except that human nature is such that the inevitable question is asked. Could I sell more with better quality, or could I save time with less poor-quality. This, I think, is the fundamental question that is actually being asked by most Dealers.
There was a post here from someone regarding Truecar "hater's" and his math seemed to show a very logical approach. You might 'feel' like you are getting screwed, but if the math says otherwise, you HAVE to ignore the emotion.
So to be succinct, I think that Dealers should worry less about the Cost-Per-Lead and focus more on the %-Leads-to-Close or TimeSpent-Per-Closed-Lead (but this deserves it's own topic of Metrics and 'how' to measure them). The reason I think this measurement is more important, is that it allows you to 'predict' the REAL cost per Lead, and helps take some of the emotion out of the conversation. Further, the Lead Gen market is saturated, the models are almost identical, the all sell that their 'mix' or 'formula' is unique, and so we need to measure performance different, since they themselves aren't able to actually differentiate.
Sorry for the blathering reply... I could talk about this topic on-and-on.
So, as a rookie member, I hope I haven't smudged, broken or stomped on any posting rules (I don't think I have

...) Or ruffled any feathers, that wasn't my intent.
JP.