Business Week published last month an excellent cover story about the subject of Click Fraud.
Click fraud is probably one of the biggest problems we face in the web world. No one can say exactly how extensive is the problem, but we are probably talking about hundreds of millions of dollars that advertisers pay for nothing.
In the heart of the problem lies the fact that it is quite easy to fake clicks. Simply pay for someone to click randomly on links in your site. I had a friend who opened a dummy web site, put some dummy content, added google ads and asked his brother to click the ads every now and than. He got about 20$ from Google every month (which mean the advertiser paid a lot more for those clicks).
The problem gets even worse when companies are using click fraud as a way to fight the competition. For example company A that competes with company B, pay people around the world to click ads of company B, thus making company B waste all of her advertising budget.
Sure, you can check IP addresses and click patterns, but in the end, there will always be people who will write a better click fraud bot engine.
The other side of the problem is the fact that neither Google, Yahoo or Microsoft explain what they are doing to fight click fraud (or how they decide to reimburse advertisers that complain about being robbed using click fraud).
Although they have a reasonable reason (stating that if they will disclose their methods, it will just help hackers take advantage of the system) for doing that, the situation simply leave the advertisers in the dark.
In the end of the day you simply don't know how much money you lose for nothing. Worse, in order to complain to Google and demand money back, advertisers need a great deal of resources to inspect and explore the raw data given by Google. Many small advertisers will never have the time or resources to do that.
So how can we solve this problem?
There is no easy answer. Lets assume that as long as we will improve fraud detection algorithms, there will be hackers that will improve fraud techniques. So the answer should not be to simply improve our systems.
Probably better is to find other ways to measure the effectiveness of advertisement. Find some ways that are harder to fraud (Note. There will always be fraud, like in any other system. Our target should be to minimize it as much as possible).
Instead of simply measure clicks and impressions, a better way will be to measure actions. Legitimate users that clicks on ads and go to a company web site will probably do some actions on that site. There will probably look for some information there.
We can measure those actions. We can determine the effectiveness of the click. We can demand more money for more effective clicks, and much less or none for clicks that didn't end in any special actions.
This will require much more participation and time from non legit users, and thus will reduce the amount of click fraud significantly.