« Will advertising shift from newspapers to the Internet? | Main | GuruNet (GRU) traffic rises, but questions remain »

Wednesday, March 02, 2005

How big a problem is click-fraud?

According to The Times of India, "A growing number of housewives, college graduates, and even working professionals across metropolitan cities are rushing to click paid Internet ads to make $100 to $200... per month". Investors in search and advertising stocks had better understand the scope of the click-fraud problem. Some thoughts:

1. Most pay-per-click (PPC) ad revenue comes from search. Over 50% of Google's revenue (numbers here) came from its own search site, and the vast majority of PPC revenue generated by Yahoo!, AskJeeves, FindWhat and LookSmart also comes from search. PPC ads on search pages are susceptible to two forms of click fraud: investors looking to boost the profits of stocks they own, and competitors trying to raise their rivals' advertising and marketing costs. Both problems are probably limited in scope, since they don't generate tangible profits for the "clicker".

2. PPC ads run by publishers are more problematic. Publishers get paid for clicks on ads on their own site, and are therefore incentivized to fraudulently pay others to click on their ads. Seth Godin thinks that click-fraud is such a serious problem that the company is "a house of cards". Like the India Times, he highlights the results of a Google-search for "earn rupees clicking ads".

3. Mechanisms are available to limit click fraud. The PPC ad companies block repeated clicks from the same IP address. (That probably isn't enough, though. StopClickFraud.com describes how fraudulent clickers limit their clicks per IP address to avoid blocking.)  Oliver Thyman argues that "mathematic formulas like Fourrier Analysis might be able to calculate the level of click fraud and see unlikely changes" in usage patterns. Crucially, Google has publicly acknowledged the severity of the click fraud problem, and has hinted that it  is working on solutions.

4. Investors need to see advertising technology trends in a wider perspective.
Pay-per-click ads have been wildly successful because they are measurable and more closely tie advertising costs to advertisers' profits. Seth Goldstein convincingly argues that ads will become more and more linked to real performance: "The future of Internet advertising is sales". Note the implication here: PPC ads will converge with affiliate marketing programs, which are entirely success-based. In a world of 100% success-based advertising, there's no room for click-fraud.

The bottom line? Click fraud might raise the search engine firms' profits in the short-run. But in the long-term, it destroys advertisers' return on investment and the "performance-based" nature of PPC ads. So PPC ad-dependent companies like Google, Yahoo!, FindWhat, LookSmart, and Mama.com will be forced to combat click fraud. In that context, it's significant that FindWhat.com claimed on its Q4 conference call that it eliminated profitable revenue because it didn't deliver underlying value to advertisers.

Posted by David Jackson on March 2, 2005 at 07:20 AM in Sub-sector: Advertising, Sub-sector: Search, ticker: ASKJ, ticker: FWHT, ticker: GOOG, ticker: LOOK, ticker: MAMA, ticker: YHOO | Permalink

TrackBack

TrackBack URL for this entry:
http://www.typepad.com/t/trackback/1948679

Listed below are links to weblogs that reference How big a problem is click-fraud?:

Comments

Another conclusion might be that the value of a click falls dramatically in the short run making payment for fraudulent clicks completely unprofitable. Google's click volume would sky rocket but overall revenue and margins would not change.

Click fraud as an industry relies on overpayment by advertisers. If companies can value TV slots based on Nielson ratings why is valuing a click so hard?

Posted by: Michael | March 2, 2005 10:10 AM

Not to undermine the seriousness of the issue being discussed but did anyone notice that the Times of India article being referred to was published on "TIMES NEWS NETWORK[ MONDAY, MAY 03, 2004 08:20:34 AM ]"?

There is a discussion on the article here:
http://www.infosyssec.com/forum/viewtopic.php?t=1496
also dated May 2004.

This brings back memories of the banner clicking monkeys but with larger implications.

Posted by: Manisha | March 3, 2005 02:40 PM

Post a comment