Search Companies Team Up Against Click Fraud

isabotage3 writes to tell us that the top three search companies, Google, Microsoft, and Yahoo, have teamed up to create an alliance to combat click fraud. The fact that these three bitter rivals can team up shows just how serious the industry has become about preserving the current online advertising boom that is currently underway. From the article: "Click fraud has attracted an increasing amount of attention amid class-action lawsuits and industry studies asserting advertisers have been collectively overcharged by more than $1 billion for bogus sales leads during the past four years. Google and Yahoo contend that those estimates are gross exaggerations generated by opportunistic lawyers and online advertising consultants hoping to cash in on the anxieties triggered by their calculations."

Comments

Logging IP Addresses

[Anonymous Coward]

Ok, I don't know exactly how the advertisers track their click throughs, but wouldn't they notice if someone got 10, 100,1,000, etc... click throughs from one or a few IP addresses? It would take some really sophisticated worm/malware to get that many machines these days to do that kind of fraud.

Yes? No? Someone who's in this business educate us, please.

Re:Logging IP Addresses

[teflaime]

Google does measure this sort of thing, to the extent they can. I have a friend who runs a lot of Google ads on his site. He said their TOS says if anyone at his IP like clicks on the ads, they don't get paid, and if they click on the ads too often, he gets booted from the program.

Re:Logging IP Addresses

[Anonymous Coward]

No, for a few reasons, the most obvious being NAT. One external IP address for ??? users behind the firewall. If your site is one of the main financial services providers, you can probably expect large numbers of users to be coming from one organization like a bank.

Lazy customers. My mom, not realizing its an ad that will charge amazon to get money, always clicks on the amazon banner when she searches for it on google (which is how she and many people find it... many people are too lazy to be bothered w/ the http://www./ thing- they just set their homepage to google and type the site into the search box.

Obviously there are clear cases where this can be avoided. If you see you are getting 100 hits a day from an IP traced back to skaterdudezRus.com and you sell skateboard accessories, well then thats obvious fraud. But if your competitor was just looking to make your day a little miserable, and just took the opportunity whenever he could to get his friends to click on your ads here and there, that is much more difficult to detect.

Spend Here, or Spend There.

[fragmentate]

The primary concern about "click-fraud" is that you're being charged for clicks that were meant to intentionally drive your costs up. In essence this means wasting money, since you can't really track who meant to click and who didn't. This adds up to tens of thousands of dollars very quickly.

The irony is that many of the companies that are uncomfortable with this medium for advertising is that they're perfectly willing to spend millions on TV and print advertising where they can't even reliably track anything. And worse, they have to hope the people that actually give a damn about their product or service are even in the market. I don't know about you, but I get my drink, go to the bathroom, and pop popcorn during commercials. With online advertising -- especially on search engines -- people only see your commercials [ads] because they were looking for something related. (I could go on a tangent here about how a clothing company will bid on keywords related to automobiles... maybe later.)

Having worked in the online advertising/marketing industry (tech sector) for over 2 years now, this problem is not easily solved. The fraudsters know all about proxies, onion routing, and a host of other tricks to drive up the costs of competitors. Then, there are those that simply think it's clever to generate traffic (on IRC we called them spammers, floodbots, etc.).

We provide our clients with click-fraud reporting using our algorithms. They're pretty accurate. But, this accuracy is based on a model, which is based on 50% hard data, and 50% conjecture. What's missing with our reporting is that Google, Yahoo!, and MSN don't give it any weight, and frankly dismiss it.

I'm hopeful that this "coming together" will help client confidence so they can move away from [nearly] untrackable advertising on TV, print, and radio. It all starts with "the big 3" -- if they're willing to assist, it's much better than a 3rd party trying to decipher 3rd party results and then have to prove it to "the big 3."

Dedicated products/service to reduce click fraud

[Anonymous Coward]

Came across this company pushing their "Clickrisk" solution to combat fraud of this type while reading the 360is Security Newsletter. here. Its under the Successful Operations section. Extract: Click fraud is expected to reach $1.1B for 2005. By 2008, this is estimated to grow to $1.6B, an increase of over 45%. (Source: Clickrisk).

Not So Difficult

[bloobamator]

Click fraud detection is not so difficult. You don't have to examine "...deep psychological processes, subtle nuances of human behavior and other considerations in the mind of the clicking person..." That's just a bunch of academic navel-gazing.

All you have to do is focus on the publishers who are profiting from click fraud. If a no-name site is getting a disproportionate amount of click traffic, then you know something is wrong. And it's only the no-name sites that commit click fraud. Large, well established websites have nothing to gain by click fraud.

Google's problem is that they do not screen their publishers. Screening would actually cost money, and would also limit their click revenue. So they let any sleazebag with a website sign up for their pay-per-click programs. That's why they have so much click fraud.

Don't forget their other approach

[The Famous Brett Wat]

It's impressive that these rivals have banded together to address click-fraud, but don't forget that Google has other tricks up its corporate sleeves. As seen here a little while back, they are also looking into "cost per action" ads, which would eliminate the fraud unless the action itself could be performed in a fraudulent manner. (Bruce Schneier mentioned it in a commentary about click fraud.)

Pay-per-click isn't the only way

[The Famous Brett Wat]

The irony is that many of the companies that are uncomfortable with this medium for advertising is that they're perfectly willing to spend millions on TV and print advertising where they can't even reliably track anything.

Woah there! You had it right in the first paragraph when you said that the problem was "being charged for clicks that were meant to intentionally drive your costs up". Now all of a sudden you're on a completely different subject: the question of whether you can measure viewer response to the ad. If you sign up for a traditional TV, print, or radio ad, you can only estimate your response rate based on market research, but you know exactly what your outgoing costs are. With pay-per-click web ads, the situation is pretty much the other way around: you get good data about user activity, but your costs can only be estimated, and are subject to escalation by fraud.

But pay-per-click isn't the only revenue model out there. Pay-per-impression is considerably less prone to fraud (it can't be easily targeted if ads are randomised), and pay-per-day returns costs to the known-in-advance state. Both of these still allow tracking of user activity.

As a small-time ad-space provider, I'd far rather be hosting this kind of fraud-proof ad. That way the ad-broker can't arbitrarily accuse me of click fraud and suspend my account. It hasn't happened to me, personally, but I'm acutely aware that it could happen at any time without notice, and this precludes me from even considering it as a reliable source of income.

What is this click fraud???

[rainman_bc]

What are these ads that people are clicking on? I haven't seen one for a very long time now...

Thank you adblock and adblock filterset.g and me blocking that google script. Life is much better without that crap anyway.

^_^

Re:Odd thing to measure anyhow

[Prof.Phreak]

If you read the article, one of the things they say is how hard it is to determine the "intent" of the person clicking the ad. Are they serious shoppers, casual browsers, or even one of those teenagers who sign up for those click-for-profit type schemes. Well duh!

I often find myself clicking on many links for sites I dislike, knowing that each click is costing them money; just 'cause I feel like it :-/

Also, opening a page in a Firefox tab, and closing the tab (without viewing it), is also a pretty nice way to run up costs for some sites. And if thousands of people do this per day...

It's a silly business model these advertisers have.

As a potential advertiser, this is keeping me out

[Presence1]

After decades building small growth companies in the software industry, I'm now building another start-up in a different high-technology area. Search based advertising would be IDEAL for our market, and our start-up company size. But the threat of click-fraud keeps us out. Here's why.

Google's CEO Eric Schmidt said about click fraud is that "there is a perfect economic solution which is to let it happen.". The idea is that the price of advertising would eventually settle to an equilibirum, discounting the average rate of click fraud. E.g., if half the clicks were fraudlent, advertisers would be willing to pay half the rate, etc.

This is a good argument, but it is fatally flawed, because there is not a steady average rate of click fraud. It might be true if the only kind of click fraud was scammers trying to drive up their own ad display revenue. This type of fraud could even out to some average, which is easily accounted for by an average discount.

However, the bigger threat is targeted click fraud -- a competitor trying to drive up my costs, or a click-farmer just happening to post ad pages focusing on my market.

These targeted attacks could easily kill us overnight, by turning 10K budget into a $100K bill, or by depleting a capped budget and taking my ads 'off the air'. Either way, it is a complete waste. Worse, this waste is completely unpredictable, I might enjoy low rates and good business, or I might be shot; it is really like Russian Roulette.

This tunable targeted ad model has awsome potential for small startup companies, where the broad image/impression advertising campaigns of major brands make no sense for them. Sadly, the click-fraud seriously poisons the well. You can see it in Google's ad revenues starting to flatten last quarter .

If Google actually solves the click-fraud problem, I'll not only use AdSense, I'll also buy their stock. I expect that many others will also start using it once it can be trusted, and their revenue will grow prodigiously. Until that, I'm using other methods.