Click-Fraud A Gathering Storm

Not much news in Wired's piece on click fraud, but a good overview of the problem and its potential to bring the search business to its knees.

The click-fraud story has been around for a while, so it is viewed as old news.  The mild hysteria that followed Google CFO George Reyes' observation a year ago that fraud could kill the company faded as, instead, the company blew the doors off quarter after quarter and the problem appeared under control. A year later, the joy of being aboard the Google rocket ride is now so entrenched and widespread--and the click-fraud story so tired--that each new piece is greeted with yawns: just another tabloid trying to whip up controversy.

And perhaps click fraud is, in fact, today's equivalent of the mid-90s "You can't make money on the Internet"--an erroneous doomsday chorus that caused countless portfolio managers to miss the best years of boom.  More likely, however, it's a gathering storm.

The Wired piece provides an excellent compendium of the flavors of the scourge: click farms (warehouses full of people paid to click all day), competitor clicks, bot clicks, zombie-network clicks, impression fraud (repeated searches of your competitor's ads without clicking until the search engine concludes the ads are ineffective and drops them).  In isolation, with careful (expensive) monitoring, all are controllable, but taken together, they add up to a major headache.  And compounding the problem is the fact that no one even knows for sure how widespread it is: the search engines aren't talking, and, truth be told, they probably don't even know.   

So, what can we conclude?  If click fraud is not growing (if the search engines have it under control), then there is no problem: It's just a cost of doing business that is already factored into ROIs.  If fraud is growing, however, which seems likely, then the impact will eventually be felt in several ways.

• Margin compression.  The more "noise" in click results, the more it will cost search vendors to filter, serve, and analyze click streams, and provide compensatory refunds.  The same price per click, in other words, will contribute less to the bottom line.

• Higher monitoring costs for advertisers leads to lower ROIs and pressure on keyword prices.  Advertisers will not be able to rely on the search engines to monitor their clicks, so they'll have to pay for software, services, and employees to do it themselves.  This will add to their costs and, therefore, hurt their ROIs.  In a rational world, keyword prices will start trending down.  (A start of this trend would include announcements like that from FTD on Friday: that online advertising costs were so high that they were no longer economically justifiable).

• Current glamor and excitement about online advertising replaced with fear, uncertainty and disgust--driving dollars back offline.  If the search engines aren't quick enough to grant discounts and refunds, or if enough advertisers conclude that even the clicks they pay for contain a significant percentage of fraud, the search business could experience the same backlash that crippled the email business back in 2000--a backlash from which it is only now recovering.

The main argument against the click-fraud threat is ROI: advertisers know exactly what they're spending and what they're getting in return, so they will never be burned by fraudulent clicks.  Assuming this is true (and some of the click-fraud horror stories suggests it isn't), all it means is that the advertisers won't get hurt: They'll just slow or stop spending on search and send their marketing dollars elsewhere.  The search engines, meanwhile, will have to increase fraud-control spending at the same time that they are losing revenue (crushing margins), and then spend another pot of money to lure advertisers back.

The other argument is that the business will eventually move to pay-per-customer or pay-per-purchase model rather than pay-per-click, thus eliminating the issue (For some discussion of this, please see the comment string here).  Such a shift would indeed solve much of the problem, but it would also require major changes in habits, data sharing, and technology, all of which would take time.  It would also eliminate (or at least complicate) the search engines' ability to get paid for delivering customers who are researching now and buying later or searchers who are honestly looking but choose not to buy (there is branding value in having someone read your ad and check out your web site, even if they don't buy something).   Bottom line, it seems there will ultimately be room for both pay-per-click and pay-per-purchase, and finding the right mix will take time.

Since there is no way to accurately monitor the level of click fraud in the system, we can only gather up the anecdotal crumbs and try to divine what they mean.  The bad news is that the anecdotes seem to be plentiful these days and that the threat is serious enough that it really could cripple the business.

(Thanks to John Battelle for linking to the Wired story...)

Comments

[John F.]

I just came across this while perusing job listings:

http://hotjobs.yahoo.com/jobseeker/jobsearch/job_detail.html;_ylt=AptIc8iKfhG1xFWzQJnx.YH6Q6IX?job_id=J6SN09UC1

So, at least they see it as a threat and are trying to do something about it...

[sigma]

"Click fraud"? You GOT'A be KIDDING!

First, for "warehouses full of people paid to click all day", is that REALLY how it would be done? I mean, why not just write some software to do the work? Warehouses full of computers? Maybe. People? No.

Second, and much more important, not so easy for Google NOT to know the internet protocol (IP) address of the computer that did the click. Sooooooooooo, just think for a milli- micro- nano- pico- femtosecond: What IP addresses look clearly like they are from warehouses full of computers clicking ASAP all day? Then send those clicks to the bit bucket. Presto: Clean data and no problem from "click fraud". Page and Brin would be tempted to throw month old toadfish at any Google programmer who didn't instantly see a way to do such filtering.

[Max Baker]

Click-fraud is the Google Killer.
It is very easy for somebody to get min $100 or $1000 per day
Google cannot do anything with it.
They just stop or close such $50000 accounts without any explanations.
Sure Google blacklisted everything suspicious and then you cannot make $1 a day with AdSence. As result, we stoped using Google Ads on PondPine.com
Also, I will never buy Google stocks for more than $90. Current target price of $600 or $2000 is a bubble.

[Network]

Great work, i like it. Thanks for sharing.
xxxecemxxx Egitim Egitim Exsohbet
Chat Sohbet Chat Sohbet
Chat Sohbet