PPC Click Fraud: 1 in 5 Is Fake, Costing $1.1B Yearly?

For a long time I've thought that click fraud in pay-per-click (PPC) advertising was a massive problem in the Web advertising business. So I was interested to see this announcement today on the TMCnet Web site, arriving via the BusinessWire feed:

Surging ''Click Fraud'' Imperils Internet Marketing; Click-fraud costs may hit $1 billion yearly

SPSS, a provider of predictive analytics software, issued this press release to promote its products as a solution for detecting click fraud. Predictive analytics, the company says, can "identify advertisements that exhibit irregular behavior patterns."

In my own Internet business (EmailResults.com), I have used PPC advertising to drive traffic, and I've noticed that some PPC services provide better-quality traffic, that is, a stream of qualified users who will respond to the offer at a reasonable rate. So for the most part I have stuck with Overture and Google Adwords and have stopped using About.com. Recently I decided to try Business.com but haven't yet come to any conclusions about the quality of the traffic they send to the site.

Here's how I imagine click fraud works: A PPC advertising network finds subcontractors who offer to provide a stream of users to the network to click on ads. The users who click on the ads are paid to do so. If you are living in Asia, Africa or South America, it could be a relatively well-paying job to sit at a computer and click on ads for a tenth of a cent per click or some such rate.

This is probably an over-simplified scenario, but I think that's basically how it works. It's probably done under-the-table and through shadow companies so it's not real easy to track down.

SPSS's news release gives five recommendations for overcoming click fraud (I quote):

1. Determine the policy for online search advertisements before signing any contracts.

2. Use Web analytics to better understand traffic to your Website.

3. Monitor Website activity closely through the entire lifespan of an online campaign.

4. Apply predictive analytics software to anticipate fraud.

5. Flag possible fraud cases with online search advertisers as soon as identified.

AB -- 7/18/05