Pitfalls of PPA Advertising

    March 26, 2007

I love it when I get a comment that is better than the original post. Last week, I chimed in with some thoughts on Google’s experiment with pay-per-action (PPA) advertising, and I got a hugely detailed comment this weekend on that post, so go re-read that entry to see it. (I’ll wait.) When you come back, I’d like to talk about it more.

TrafficSentry comments










In my original post, I mentioned that I expected return fraud to be far less of a problem than click fraud has been, but ClickHawk from TrafficSentry believes return fraud will be zero. He (I don’t actually know if it is a "he" or a "she") may well know more about that than me. He writes, "’Return Fraud’ is a non-issue. Not only does the time/return ratio not make sense for would-be perpetrators, but all existing PPA programs already have built-in measures to debit affiliate accounts upon returns. In fact, this very feature opens up a completely new type of fraud; Advertiser Fraud (merchant fraud against publishers) can occur when unscrupulous advertisers falsify return records in order to receive advertising refunds."

My speculation centers mainly around what kind of relationship Google wants with its advertisers. Return fraud is near zero for affiliate marketers because the product is being returned to the merchant—the one who pays the affiliate. With Google’s program, the return would be made to the affiliate/advertiser, so I am not sure how Google could verify that it happened.

As ClickHawk points out, if Google merely takes the advertiser’s word for the returns, then it is exposed to a different danger, a fraud perpetrated on Google by the advertiser claiming fictitious returns. To me, that is the most interesting part of how Google structures its program, and I have not seen anything written on this yet. Perhaps my commenter or others know of how other pay per action programs work, but I have not found any public information.

ClickHawk goes on to make an even more interesting point about whether the supply of PPA ad space will be available: "PPA is a boon for the advertiser. Not only does it wipe out the possibility of fraud, but it also completely removes his risk, as he is only paying for sales. Now, the publisher is taking all of the risk, and has no idea what, if any, revenue he will receive for the advertising. If the only risk was proving that his audience was a good enough match to produce good response to the ad, the publisher would be more likely to accept the terms (this is, in effect, CPC). However, the publisher is also taking the gamble that the advertiser has both a marketable product AND an effective marketing plan, all the way through the conversion process to the close of the sale. All of this is outside of the publisher’s control, so that’s quite a big gamble. If their is little market for the advertiser’s product, or if the advertisers price is not competitive, or the advertiser’s web site is unprofessional in appearance, or any number of other reasons that lead to loss of the sale, the publisher gets nothing."

To me, this is something we’ll just have to see play out. We have very little information on the nuts and bolts of Google’s program yet, but Google might handle the performance problem (and thus the supply problem) by applying the same kind of arithmetic to Google PPA ads that PPC ads get.

Although Google is now dabbling with quality factors in its PPC ranking algorithm, the original bid and click rate approach essentially ranks ads at the top that pay Google more. The original Goto.com ranking method merely assigned the top bidder the top spot, so the risk was that bidders used ads that did not get clicked enough, but had very high bids—Goto had to "suspend" ads with low click rates to ameliorate this danger. Google eliminated that problem completely by adding click rate to the formula.

I see no reason why Google can’t do the same with PPA. The feedback loop is more complex, but Google Analytics allows conversions (actions) to be tracked and Google will know how much each ad returns to its coffers. Google could rank PPA ads alongside PPC ads based on which is returning the most money. Doing so would ensure that PPA ads must perform to the level of PPC ads (from the standpoint of Google’s return) and would allocate supply in the exact ratio that performance dictated.

We’ll need to get more information about how the program works before we know what approach Google is taking—they are smart folks, so they have probably come up with something better than me. Google has an opportunity to integrate PPA into its successful PPC approach, and it would be big news if it did.