John Battelle's Searchblog

A worthy read. Lite on answers, but it did help me understand the complexities of the problem better, and a sense that it may be much more widespread than I had thought. I did not quite see how Google Wallet would prevent this, but snap.com's approach seemed like a good one to me.

Otherwise, why not just charge businesses for the first click from a certain computer, say each week. If that computer clicks more than once on an ad, there is no extra charge; just up the fee for the first click some. One a week of click fraud ain't so bad. Seems that would help a little.

Google does install a pervasive cookie in temp internet files - surely substantial amounts of information can be retrieved from those stats.

Also, SEOs, SEMs and Webmasters should be installing qualit client side stats/trackers on a seperate Landing page put up just for the PPCs, and evaluating the data frequently.

If naive Web site owners are attempting Search Engine Marketing on their own - they are probably very easy prey. If they are a small business, they may not even get high quality service from the Search Engines Campaign set-up sales department - who may be working on some sort of commission or incentive program.

There are many current stories about PPC Class Action lawsuits in progress that also offer valuable info not included in that fine article.

in orwell's chilling fntasy, when the pigs change one of the seven commandments from "all animals are created equal" to "all animals are created equal but some animals are more equal than others" the other beasts on the farm slowly comprehend the ambitious intent.

and when a corporation is faced with the prospect of needing to maintain growth even after it has swelled to a 100b$+ concern, comprimises simply must be made.

"do no evil" now begs the question,

what is evil?

he he he...

Google does track IP numbers that click their ads, this has happened from the very beginning. They don't discuss their efforts to curb click-fraud, but you can be sure that fraud will always have a profile different than real customer clicks.

A massive global attack against the entire system might be a danger, but that would be more about terrorism than click-fraud.

A close monitoring will allow an AdWords advertiser to turn off the campaingn when odd patterns develop. Too many people will just sign up for the program and not check back on it until the bills arrive, that is the problem. It has to be watched, if you dilligently present logs of suspected fraud you won't get burned.

Guarding against repeated clicks from the same computer or IP will only protect against the most trivial types of click fraud. Here's a simple way of circumventing such checks:

Let's say somebody offers a cool computer game for free download. (Or a free porn site, or free music downloads, or some other combination of services that appeal to a broad audience and yet can be offered to millions with little extra cost.)

In return for the free service, all you have to do is to agree to have a small program run in the background on your machine on an intermittent basis when it's idle anyway. No harm to you. (And most people wouldn't know that they have "opted-in" to this anyway, because we know from countless user studies that people don't read the agreements they click "I agree" to.)

With a compelling offer, you can easily get 10 million computers around the world in your network. And they are not "zombies" because the user has opted in and because you are doing nothing bad to the user.

Once a day (or week, or whatever), your software connects to a central server and downloads a list of keywords to search for and websites to harrass. At random intervals, the software connects to any of a number of search engines and issues a number of searches. Most of them for queries that are not targeted, but simply using keywords that are in the news or found on a random website. (This to throw off any analysis of whether an IP number is predominantly searching for targeted keywords.) On a few percent of these searches, the software clicks on a random ad - again to foil any analysis of systematic click on targeted ads.

Once or twice per day, the software issues a targeted query and clicks on one of the targeted ads. (Randomly selecting the query from the thousands that were downloaded.)

With ten million computers in the network, it should be possible to issue fifteen million fraudulent clicks per day, which will be enough to bring quite a lot of targeted campaigns to their knees.

The key to making this happen is to be random in all behaviors, and to burry the targeted clicks in a bunch of other clicks that are all over the map. (However, don't be so random as to be identifiable random - do most of the non-targeted queries on things that lots of people are searching for. Also, don't run the software 24 hours per day, but emulate normal user login patterns.)

As quickly as one side can write pattern recognizers to screen out the fraudulent clicks, it's possible for the other side to invent new randomizing factors. The real problem is to charge money for something that's not a full commitment on the part of the user.

Ultimately, we will probably move from pay-per-click to pay-per-action. This is the way Amazon's affiliate program has worked from the beginning, so they probably hold lots of patents in the area.

Google actively blocks the above "spyware", yahoo goes out of their way to recruit companies like that. MOst of the recruiting is done by yahoo reselling the feed to a third party. The third party then gives these spyware vendors etc a quota per day and the third party then cloaks the referal string etc and mix in some high converting traffic to make it look like legit traffic.

I know some people doing this and they are doing $ millions a month, they got banned from google within 30 days but not yahoo. Note that the traffic isn't fraud, its just low converting popup spam etc.

Yahoo needs to do some serious house cleaning if they ever expect to compete with adsense.

For some data on click fraud - see
Click Fraud by Brendan Kitts, Benjamin LeBlanc, Ryan Meech and Parameshvyas Laxminarayan

Bulletin of the American Society for Information Science and Technology

http://www.asis.org/Bulletin/Dec-05/clickfraud.html

The authors show that the search engines may already be accounting for the percentage of clicks that are fraud, and in the end, it may all even out.

Seeing as the issue of click fraud (and other types of ad fraud) have been debated for over ten years now, and there's still no solution, perhaps it's time to admit there isn't going to be one (at least as long as companies still insist on charging for clicks, or anything else that's easily defrauded).

Google does track IP numbers that click their ads, this has happened from the very beginning. They don't discuss their efforts to curb click-fraud, but you can be sure that fraud will always have a profile different than real customer clicks.

Google does install a pervasive cookie in temp internet files - surely substantial amounts of information can be retrieved from those stats...