Google tests an ad idea: if you don’t get results, don’t pay

The vast majority of advertisers now pay Google when a user clicks on ads that are displayed alongside its search results or on other Web sites, while some are billed based on how many people view the ads.

“We’re optimistic that it will be something that will be very compelling for advertisers,” said Susan Wojcicki, vice president of product management at Google. Wojcicki said the system would also give participating Web publishers a wider choice of ad types for their sites.

Under the “cost per action” system, advertisers decide what they are willing to pay for a specific action, like a purchase or a software download. Armed with that information, Web site publishers then choose whether to run a specific ad or group of ads on their sites.

Many advertisers find cost-per-action appealing, as it greatly reduces their risk, since they are not charged for ads that are ineffective. The model has long been used online by “affiliate marketing” companies like ValueClick, which have created networks of hundreds or thousands of Web sites that display small ads for e-commerce sites. The publishers are paid when they refer a user who makes a purchase.

But many other companies are using cost-per-action ads in different ways. “We think it is a model that all the large players in search will be embracing over time,” said Tom McGovern, the chief executive of Snap.

For the time being, Google is not putting cost-per-action ads next to search results, limiting them to publishers’ Web sites and essentially creating its own affiliate marketing network. Industry insiders said Google’s entry into the market was likely to accelerate its growth.

“This is a big market at an early stage,” said Ellen Siminoff, chief executive of Efficient Frontier, a search marketing firm.

Cost-per-action ads have another advantage: They virtually eliminate the problem of click fraud, a scam in which people or computers generate clicks on ads for the sole purpose of getting a payment.