I've had some experience in the distant past running some reg-path sites and the number of people who go through the whole process is very low. Maybe not as low as the percentage you quoted, but significantly below 1%. The number is so low that the cost of the item spread across all users is negligible. However, if you are dealing with as much traffic as you state, you will have people go through your whole process. So you have to promise something you will actually deliver. Otherwise there will be complaints and if there are enough, the AG might come sniffing around.
One thing I can tell you from experience is that keeping a reg-path site as profitable as possible is a ton of work. It requires constant maintenance - from recruiting new partners, to campaign management within the reg-path flow (setting up new offers, pulling down old ones, cap management, etc.), to invoicing and other accounts receivable pleasantries. Sure, you could just make deals with a couple agencies/networks, but you'll be trading off rates to save you on recruitment/payment work. You'll still be faced with all the campaign management issues. If you don't do all this work, your effective per-lead rate will fall significantly below what you can get from existing e-mail submit merchants. Even finely tuned, it'll be a lot of work for you to get your own reg-path to generate an effective per-lead rate that is higher than what you can get from existing email submit merchants.
Considering that, you have to ask yourself if doing all that extra work is worth the benefit of the regular hassles associated with running an e-mail submit campaign.
Note that my comments above are based on the fact that you are simply looking to build something for your own in-house use. If you are ever looking to open that up to affiliates/networks, you might find it certainly worth the effort.