Log in

No account? Create an account

Previous Entry | Next Entry

Google testing CPA ads

SeekingAlpha reports that Google is testing CPA (cost per action) ads. The action may be a sale or it may be a lead (e.g. a registration on a web site). The publisher of the ad would take a percentage of the revenue upon an action. (The rest would go to Google.)

CPA is subject to fraud, as is CPC and CPM. For example, software can be easily programmed to generate fake registrations (ask anyone whose blog has been spammed lately). There are ways to limit the exposure to fraud, such as requiring (difficult to guess) passwords or captchas. It's even more difficult to defraud CPA when payment is made to the publisher upon a sale (although it is subject credit card fraud and the like). While I personally think that fixed fees are the least subject to fraud of all the ad payment business models, if advertisers just feel they're too much at risk of paying for insufficient traffic with a fixed fee, CPA is less risky than CPC and CPM.

There's been quite a debate about CPA in several blogs and forums. A common concern among publishers is that they stand to make less money through CPA, in particular due to shady merchants who are unwilling to pay for sales, leads, etc. There seems to be some evidence that this is already happening in some affiliate programs such as Commission Junction. While this is certainly a problem, I think the locus of fraud-incurred damage is much easier to contain and manage with CPA than with CPC or CPM. Also, there are legal mechanisms to deal with merchants who refuse to pay, whereas the (il)legality of click fraud is still at question. (For example, is someone who innocently clicks repeatedly on an ad guilty of click fraud? Maybe, maybe not. But the traffic pattern generated will be similar (possibly even identical) to patterns identified as fraudulent.

In the grand scheme of things, there's no such thing as a system of nontrivial complexity that is incapable of being compromised. The goal is to minimize risk. So the debate will continue to rage on ...