We're often asked why we charge on a cost-per-click basis instead of taking a cut of revenue based on a view-through attribution model as so many (but not all) of our competitors do. What's wrong with CPA based on view-throughs with a 30-day look-back? Well, there's nothing inherently wrong with that model--any model can work for client and vendor if you can measure the right things--but for most of our clients, it's not the best choice.
There are plenty of articles on the potential pitfalls and challenges of view-through tracking. For a quick primer, read this article on CMO.com: Different Views of View-Through Tracking.
The problem with view-through attribution is that most advertisers aren't set up to measure the real impact those views have on sales. In our experience it is unlikely that the advertiser is doing sufficient modeling to truly understand how many of those views are really leading to conversions. As the article on CMO.com notes, just because an ad was served to a customer does not mean they even saw it. In fact, some less-scrupulous CPA companies buy a large volume of very cheap, below-the fold ads that are never seen, because it puts them in the conversion path and they get paid on the view-through conversion despite having nothing whatsoever to do with the sale. Some of these ads are even tiny little text links at the bottom of pages that have no real value.
If an advertiser sets up decent modeling, where they are tracking all the marketing messages a consumer is exposed to, or performing lift analysis with control groups, and their contract with a vendor recognizes these models so that they pay where real value is delivered, a CPA model using view-through data may be attractive. Most of our clients do not have the resources or expertise to perform that analysis, however. So a CPC model makes more sense.
When a click is the relevant action, you eliminate a lot of uncertainty. You know the consumer engaged with the ad and visited the client site as a result. The ad position is no longer an issue, and there's no longer a question as to whether the consumer saw and was influenced to some degree by the ad. The question now is simply how many of those people convert, and whether the overall return on ad spend across all marketing efforts is working for the client. (That, by the way, is a better way to look at attribution and return than 'last-click' attribution, which unrealistically assumes that a consumer was only influenced by the very last interaction with the advertiser, not by all of them.) We still provide our clients with all the data they need on impressions and view-through conversions to build into their modeling, we just don't charge them based on that data.
Using CPC is, for most advertisers, the clearest relationship between marketing dollars spent and value received.