Defining Clicks and Click Fraud

abb_road writes "Google, Microsoft and Yahoo have banded together and created the Click Measurement Group, with the goal of creating a standard definition for a 'click'. The group will have some access to the three companies' click data, although the access won't be unlimited. The move comes in response to advertisers who claim that click fraud is costing them almost $1 billion dollars a year, and who have hit Google and Yahoo with lawsuits alleging negligence in fighting click fraud."

I *like* dupes.

[WilliamSChips]

I never caught this the first time around.

Signal to Noise

[gartogg]

All three companies involved do so because it is beneficial to their image to be seen fighting click fraud, even though they know that the recommendations for eliminating it will be useless - it's a technology battle, and it will escalate. No matter what the recommendations are, they won't do much to stop persistant abusers of the system. But, on the other side , it doesn't matter.

The real solution here, as usual, is the free market. Advertisers will decide where to spend their ad budget, and they either think this is a problem or not. The solution will boil down to convincing these (probably technologically savvy) ad people to buy ads. That's why having a standard is useful - it looks good. And judging by Google's profits, corporate wallets are voting yes to online ads. If click fraud was a real problem, they wouldn't.

Oh, this is great....

[duplicate-nickname]

I can guarantee that their definition of a "click" is going to favor the bottom line of Google/Yahoo/MSN and make it even harder for advertisers to get their money back on fraudulent clicks.

Very interesting and difficult problem

[poliopteragriseoapte]

Defining what counts as a click is a very interesting and difficult problem. The main trouble is that you have to get a definition such that, even if the click-spammers know it, they cannot take advantage from the knowledge.

I have the impression that right now, click fraud is fought using statistical criteria to identify real and fake clicks. If you publish a definition of what is a real click, the definition has to be very good and clever, so that fraudsters cannot simply write code that generates fake clicks that satisfy the definition.

Re:Might not be exactly the same...

[gartogg]

This article is actually an interview with someone involved, and brings up substantially more issues than the previous post.

Not that Zonk, out wonderful poster, noticed this, as we can see from the fact that it isn't mentioned in the post.

Re:Oh, this is great....

[quarrel]

I'd agree with you. However, with the rate online advertising is growing, and the amount of money they're likely to gain from that growth, the best thing for their bottom line may well be to keep all of this legit and above board.

Why risk the goose that lays your golden eggs?

--Q

Oh FFS

[Greyfox]

If it's that much of a problem, work out on average how many (legitimate) clicks generate a sales lead and pay that much more for actual sales leads generated instead of per-click. Then the whole problem of click fraud goes away (And gets replaced by sales lead fraud.)

Re:The solution is easy

[joshetc]

I think there would be even more problems with fruad paying per impression. Its extremely simple to make an advertisement invisible on a website. While being paid per impression webmasters have no incentive to actually let visitors see the advertisements. Which is why I suggest some sort of minimum impression / click ratio. The cost of 50 clicks guaruntees you 50 clicks AND 1000 impressions or whatever a decent click through ratio might be.

This way no website can have a 100% click through ratio, and if they did the companies would only be charged for 5% and the website would only be paid for 5% CTR.