Searchonomics

Yes, we're talking about the economics of search, something very few people take the time to understand. Many people scoffed at Google's business model as it was on its way to be a multi-billion-dollar juggernaut. As of late people have begun to respect and fear Google appropriately. However, there has been some chatter about the vulnerability of the pay-per-click model on the internet. Microsoft's recent purchase of JellyFish and, more importantly, its foray into a cost-per-action paid search model (let's forget about the cash rebates for searchers for the moment) has people talking about a potential paradigm shift in paid search. At the root of this is the perception that click fraud is such a threat to advertisers that they will suddenly flock to whichever search engine comes up with a viable cost-per-action service. Let's discuss this notion.

Ads follow eyeballs

First of all, an obvious lesson that analysts often forget is that advertisers follow audiences first and foremost. If Google continues to widen its lead in search on a query basis, Google will also continue to dominate in paid search on a revenue basis.

Click Fraud

Now, speaking more directly to click fraud, it is a threat to the pay-per-click model, but not to the extent that people often think. Although I have just as strong an aversion as the next person to people making and losing money on fraudulent clicks, the pay-per-click model naturally adjusts in the long term to account for these bogus transactions.

For example, if an advertiser suffers a series of fraudulent clicks, either Google will refund the advertiser her money, or the ROI for that advertiser goes down. In the former case, no harm is done. In the latter case, the advertiser will likely bid less money for the clicks that no longer offer the same utility, and her ROI will likely trend back to where it was previously.

Would it be better to just charge advertisers only on converted sales?

Yes and no, depending on your perspective. On a theoretical level (we are talking about economics here) a cost-per-sale model would indeed be a better economic model. If all advertisers pay exactly how much a sale is worth to them, then those advertisers would have their marginal benefit equal to their marginal costs, their economic profits (as opposed to accounting profits, which are different) would be zero, and the search engine would capture all of the possible value from the advertiser. This scenario makes economists really happy.

This theoretical notion is backed up in the real world by statements made by many advertisers who said they would likely pay more on a per-sale basis if the cost for that sale were fixed and guaranteed.

Then, what's the holdup?

It seems that everyone makes out better with this type of system. Right? Well, not really. In reality, the paid search marketplace is much more complex, with some advertisers wasting large sums of money on keywords and others practically printing money by converting sales for pennies on the dollar. For the former type, a cost-per-action (action being any meaningful action such as an e-mail sign-up, a sales lead, or an actual sale) system will make ad campaigns much easier to manage and much more lucrative. However, for the latter type, the removal of the less accountable pay-per-click and CPM payment models would be a huge threat to the profitability of savvy advertisers.

Many e-commerce sites have made it their business to actively manage keyword campaigns to maximize ROI. A cost-per-action system would, in aggregate, increase these successful firms' ad ratios, threatening the sustainability of their business. Some firms will go from posting great economic and accounting profits to going out of business very quickly. So, not everyone will benefit from a cost-per-action model that essentially levels the advertising playing field.

Oddly enough, the leveling of the playing field in this case will likely benefit deep-pocketed advertisers more so than smaller advertisers. For many keyword phrases the overall prices will rise as a result of a more accountable advertising system. Larger companies who pay high prices on generic keyword phrases for defensive purposes will continue to do so, but these companies might also find it worthwhile to seek a broader array of keywords if they are only paying for a pre-defined action that has calculable value. This will likely cause a raise in costs for many small businesses that rely on the low-hanging fruit that long-tail keywords have provided for quite some time. Also, the conversion rates for these e-commerce sites will no longer be the reward that it used to, as a sale would then likely cost as much as its incremental benefit, and not a small fraction of it, as is often the case for a savvy entrepreneur.

What are the implications for Google and Microsoft?

Even though Google stands to capture more value from advertisers by switching to a more accountable ad system, the culture at Google to this point has largely been shaped by the pay-per-click model and the small businesses who utilize the system. Since so much of Google's revenue comes from small businesses, the company might be hesitant to uproot the ad system that has been so good to it. The evidence: Google could very easily implement such a system, but hasn't yet. Google already allows you to track conversions on your site for a number of different goals (actions). Applying the payment system to reflect these statistics is not a stretch at all for the world's most powerful tech company. (Google has done limited testing of a cost-per-action ad system with select customers.)

Microsoft's new cost-per-action system in conjunction with cash incentives is an interesting move because it strives to innovate with its ad system and attract new search users at the same time. With that said, it doesn't present a solution to search itself and in the long term undermines Microsoft's clout in web search by turning it into a product search engine. (For more about this, refer back to my post "It's Search, Stupid.")