article [today.reuters.co.uk]
"A seller of online marketing tools said on Wednesday it sued Google Inc., charging that the Web search giant has failed to protect users of its advertising program from "click fraud," costing them at least $5 million."
Doesn't surprise me this finally happened. They are going for a class action suit.
[edited by: Woz at 11:32 am (utc) on July 1, 2005]
[edit reason] Fixed Scrollism [/edit]
click fraud rates of as high as 38 percent.
I can see why the formula worked. Of course, this statistic seems a little bloated to me and I know that click fraud is extremely market sensitive. Is anyone here able to validate Click Defense’s claim of such a high percentage of click fraud?
Let me play devil’s advocate......
It does seem very suspect that the plaintiff in this suit stands to gain enormous publicity regardless of whether they win or lose, but click fraud isn’t exactly hurting Google either.
Public opinion is the law with any e-commerce or Internet business. Right now Google and other providers are managing to keep click fraud under the radar of your average stockholder and still collect the 10-30% (or whatever) of revenue that click fraud creates. Please correct me if I’m wrong, but until click fraud starts to generate negative publicity in-turn hurting stock value, it appears to be their best seller. John Doe stockholder has no idea what technology is behind any of this and could care less what an API is. They want to see stock prices climb. At this point click fraud is making that happen. When mainstream media blows click fraud out of proportion stock prices will hit the floor and Google and every other provider will react accordingly.
Until then the only thing worth wasting time and energy on is making sure your ROI is positive regardless of click fraud. If the advertising is working, go with it. If it’s not, drop it.
We started using a click reporting software about a month ago on one of our most competitive keywords. So far, I only show about 3 instances where we've had more than 5 clicks from the same IP address. We've never had more than 7 clicks from any individual IP. Sure, we're not bidding on $50 keywords, but our average CPC on this word runs about $2 to $3 and brings in several hundred sessions per day, so it is somewhat competitive.
So, I would like to hear from people using this forum;
1. What is the best way to identify click fraud? Click spikes? Conversion dips? IP addresses? Session logs?
2. What percentage of your traffic do you truly think is click fraud, and what data do you have to support it?
3. What experiences have you had in confronting the major search engines requests for refunds?
My personal opinion is that about 5% to 10% of the traffic my site receives is fraudulent, but I have no data to substantiate it, and the numbers that I do receive are so low and inconsistent that I am unable to make any sort of a case for a refund. I am sure that there are competitors that have some very advanced click fraud techniques that probably try to add an additional 10% to 15% more traffic to our site just to make us barely unprofitable, but how do you find this?
You are Yahoo and Google's best friend. Actually we all are. They know there minimally a very small % of click fraud that affects all advertisers at some point. And you know what? Those 'suspicious' clicks mean the same revue to them as legitimate clicks.
The key there is unmonitored syndication. Most of the click fraud out there isn't competitive click fraud. It's very easy to track and filter out. A recent marketing experiments report tried to commit competitive click fraud on itself and had only about a 10% effectiveness rate. Competitive click fraud follows a few easily discernible patterns and beyond radically changing one's behavior, one isn't able to really hurt a competitior that badly.
The much more dangerous form of click fraud is automated or bot-based click fraud. There are bots currently in existence which can grab proxy servers for unique IPs, come in as IE 6.0, spend time on the search page, click on paid links at a certain CTR and then establish click depth on advertisers' sites so that for all they know, they just didn't close the deal.
I think the 38% is probably high on general traffic, but there are plenty of ways to hop skip and jump through Google's (or Yahoo's or Findwhat's or Enhance's) defenses that are in the grasp of most of the webmasters on this board.
Given that such accessible methods exist, one needs only follow the incentives to see where click fraud will be prominent. And Google AdSense offers the most money of all to most site owners. While our syndicated Overture feed typically beats AdSense performance for website owners, our network is small and we monitor all attempts at click fraud in real time, so we never pass the clicks to the advertiser at all - they get them for free.
If people, however, can game Google for a few hundred extra bucks a month, they're going to do it, regardless of the fact that it's unethical. Should Google be held responsible for allowing this to continue? That's a tough question. Our technology prevents the nonsense I've described above, but we rolled out from the get-go.
Google, on the other hand is sort of in a corner. Their interest is aligned with the fraudsters'. If they stopped all fraud right now and their revenue dropped say, 10-15% (which is about what I expect on their entire network, 95% of that from AdSense syndication), their magestic growth which has so captured Wall Street's imagination will flatten significantly in the short term, which could do horrible things to their stock price.
If they do stop it immediately, they'll also have hordes of lawsuits dropped on them, much bigger and much more legitimate than this one, because they'll basically be admitting to a frankly disturbing level of fraudulent revenue.
If they don't stop it, they'll gradually lose the revenue on the AdSense side of things (47% of gross revenue last quarter) becaus advertisers will opt out of AdSense display more than they already do.
Their only choice is to placate advertisers and softly and gently reduce the CF on their network if they can figure out how to do so so that growth isn't slowed too much and advertisers don't bail too quickly.
The problem is 100% the ineffectiveness of the internet medium to accurately identify the theives, bandits and evil doers due to the shoddy protocol we live with but you would prefer to heap blame on the companies doing the best they can to deliver one of the best products possible on a horribly shaky foundation.
Assuming what you say is true, it should follow that the SEs should change to a business model that is not so easy to defraud, such as fixed fee ads.
Assuming what you say is true, it should follow that the SEs should change to a business model that is not so easy to defraud, such as fixed fee ads.
That's the previous CPM business model and it has all of it's own problems which is why everyone went to CPC - customers demanded it as it was more cost effective. Everyone surfing the site burns impressions with CPM and a team of people working on a web site will deplete advertiser CPM inventory, no joy there.
CPA (affiliate programs) are the safest for the advertiser but the internet technology once again ends up defrauding the PUBLISHER this time with cookies discarded, washed or worse, not the advertiser. Plus products like Norton Firewall block all the ads putting the publisher and advertiser at a disadvantage.
So CPC is the happy medium most people agree on as the fairest way to compensate third parties for advertising.
That's the previous CPM business model
Not necessarily, it could consist of something new. i.e. mentioned before for example would be a system based on bid amounts per day. Impressions nor clicks would be relevant and thus alleviate attempts at fraud based on these factors.
With fixed fees, there is much less chance of fraud. Running up excessive impressions or clicks does not affect how much money you pay. In fact, you can even verify your placement without fear that the publisher will remove you from the program for generating impressions or clicks.
I spoke with a tech support there and he doesnt sound like he knew what he was talking about. He said that Mesolethemia is a $90 keyword, and he extracted his theory from there. LOL
They dont even know what the real bids of those keywords are, they could be the suggested bids.
Click Defense doesnt have a clue and i'd assume they are struggling to get business.
I bet they are doing this to make a buzz to be known.
What a cheap shot.
As i said i spoke with the Tech Support while he was logged in to Google Adwords and checked the Mesolethemia keyword is a $90 bid. He then made a rationale from this data that it's a target of click fraud.
I think Google should block this company from advertising because they are just using Adwords for its evil purpose.
Several posts in this thread indicate that people believe Googles interests are aligned with click fraudsters. I don't think anything could be further from the truth for two relatively simple reasons. First, assuming the market for keywords is economically effecient, removal of all fraudlent clicks clicks would result in advertiser conversion rates rising by the exact same proportion as the amount of click fraud that was eliminated. Efficiency says that the advertisers who compete with other would raise there bids corresponding to the increased conversion rate. As a result the advertisers would spend the same amount of money as before, but Google would make more. So the money click fraudsters earn now is really coming from Googles potential reveune.
Secondly, any lack of confidence in Googles system inspired by click fraud inevitably leads to fewer adverstisers spending less dollars than would be spent on a system that adverstisers were 100% confident with.
I have no idea how successeful Google is at stopping click fraud, but I am certain that they have a strong incentive to try as hard as possible.
Regards
[money.cnn.com...]
Assuming 38% of CD's advertisng budget was $5mil (the loss being recouped), then they spent about $12mil on adveritising in Google over whatever period.
It's been stated all over the place, including that CNN article, that about 20% is the figure given on click-fraud (and that was last year - it's probably up).
If the above is correct, then assuming CD haven't been refunded at least $2.5mil (20% of $12mil) from Google, then they do probably deserve to win, since it would be about that figure minimum they lost. Of course if they can validate their 38%, then they deserve all of it.
I do think that bid rates would eventually rise to their current fraud-based rates, as ROI would be better and the market system would in theory allow it, but it would take time.
Google's biggest issue as it relates to this is that it is obscenely overvalued on WallStreet. If they do anything to reduce click fraud, they will, in the short term, reduce revenue growth. Let's say Google's network wide CF rate is 10-15%, as I suspect it is (80-90% of that is on GAS, by the way).
That means that 10-15% of Google's revenue is fraudulent. In that scenario either a) they don't know it and can't do anything about it, which is bad, b) they know and can't do anything about it, which is worse, or c) they know and they choose not to face the consequences of fixing it.
In any of the above cases, Google's stock price is in for a serious bruising. Most people with Google stock can feel it coming, I suppose, but as PT Barnum didn't say, there's a sucker born every minute, so there's always somebody there to pay a little more.
I do think that bid rates would eventually rise to their current fraud-based rates, as ROI would be better and the market system would in theory allow it, but it would take time.
Google's biggest issue as it relates to this is that it is obscenely overvalued on WallStreet. If they do anything to reduce click fraud, they will, in the short term, reduce revenue growth. Let's say Google's network wide CF rate is 10-15%, as I suspect it is (80-90% of that is on GAS, by the way).
That means that 10-15% of Google's revenue is fraudulent. In that scenario either a) they don't know it and can't do anything about it, which is bad, b) they know and can't do anything about it, which is worse, or c) they know and they choose not to face the consequences of fixing it.
In any of the above cases, Google's stock price is in for a serious bruising. Most people with Google stock can feel it coming, I suppose, but as PT Barnum didn't say, there's a sucker born every minute, so there's always somebody there to pay a little more."
Best post on this discussion yet. I said everything that is stated here; although this is much more professional sounding.
that is 1 billion dollars in profits last year (unconfirmed of course) of which about 90% is made via adwords, so that is
$900,000,000 that adwords generated.
taking 20% click fraud into account that means that last year
$180,000,000 of the profits that google earned were by fraudlent clicks.
as a business owner myself, that makes me sick.
a lot of people are saying that you have to blame the fraud people 100%, and that none of the blame lies on google.
let me ask you this, do you think the tech people who are employed by google are idiots? of course not, im sure they have the brightest minds in the world working for them. so they have the means, and financial power to develop anti fraud techniques that could cut out, im guessing at least 90% of the fraud that goes untouched every year.
but asking them to cut out $162 million dollars in profits is like asking a kid to give away his candy and not get anything back.
it aint gonna happen.
i for one hope the lawsuit is succesful, and that some good comes out of it, for the common man, not the publicity stunting company that is running it.
let me ask you this, do you think the tech people who are employed by google are idiots? of course not, im sure they have the brightest minds in the world working for them. so they have the means, and financial power to develop anti fraud techniques that could cut out, im guessing at least 90% of the fraud that goes untouched every year.
All click fraud cannot be tracked. When the fraudsters are employed to act like "legitimate" surfers, or use (or co-opt) machines that can simulate "legitimate" surfing, they're indistinguishable from Joe or Jane Public. There can't even be a guess as to what % of all click fraud falls into this category, because it can't be detected. Thus, it is impossible to assess what % of click fraud Google (or anyone else) can eliminate.
You're correct that some click fraud can't be monitored, at least until we can effectively read minds through some sort of strange biometric keyboard analysis trick, but it's not as hard to monitor the sophisticated click fraud that gets by Google and Findwhat and Enhance and the rest as you would suspect.
We employ our own click fraud filter technology on our distribution network. Because it works, we're able to insulate our feed provider and its 150,000 advertisers from any possibly fraudlent clicks. Because we handpick our distribution partners according to specific criteria, we don't really get any attempted click fraud from them, but what's interesting is that we can see secondary clicks from bots that have come to their site.
Let's say one of our partners buys an ad on AdSense. Many of the bots that are used in click fraud click once on the GAS ad and once again on the landing page, thus creating click depth. Well, we can see those secondary clicks that our partners are paying for. We don't have as much control to identify them when we're not looking at the first click as well, but there are definitely signs that indicate that a certain percentage of the traffic coming through google is bot based.
That's where my estimate came from.
no, its pretty easy, if you have an ip accessing google, i dunno, 5,000 times a day, then odds are, it is a bot, or automated surfer, even if it is searching random, predefined terms and clicking on them. because no human is going to sit and surf google like that, and thats just off the top of my head.
This type of click fraud is easy to detect, and is already being detected by most SEs and ad networks. The types I'm describing are far more subtle because the bots act the same way that people do who "legitimately" use the web (or are people hired or otherwise engaged for illicit purposes).
You're correct that some click fraud can't be monitored, at least until we can effectively read minds through some sort of strange biometric keyboard analysis trick, but it's not as hard to monitor the sophisticated click fraud that gets by Google and Findwhat and Enhance and the rest as you would suspect.
I never said that. I said that it was hard to detect click fraud that is designed to emulate "legitimate" human behavior, or is done by humans with fraudulent intent. I also said that since it is indistinguishable from "legitimate" human behavior, one cannot tell just how much of it is "illegitimate."
One can estimate; one cannot tell.
When you have all the click and impression data, as the SE's do, and as we do, it is *possible to detect and filter nearly all of the bot based clicks. We are doing it right now.
If you're simply monitoring on the backend based on pattern recognition, it's really hard to detect all but the most obvious attempts at CF. The SE's have to be on board, or it is pretty much useless.
You put a small piece of code on every page of your site. You slightly modify your ppc click thru URL and thats it. At the end of every month their attorneys file with the Click fraud departments at google and yahoo presenting the corresponding data that their system has collected. On average they said they collect 15% - 20% per month of your overall budget usually paid within 7 days of the first of the following month. Their cut for their services is 15% of the amount they get you refunded each month.
I have to say I am definitely interested because my overall monthly spend is significant and this 15% - 20% could be a noticeable amount for me...
do u have contact information on one? please also give me your email address, i would like to contact you after youve had time to try it out to see if its worthwhile.
that is, if thats ok with u.
I am very much so looking forward to seeing more about these companies that file complaints / suits against Google for click fraud. This is probably where the demise of Google is going to come from, and rightly so since they don't seem to have their socks pulled up all the way. But in their defense, tracking click fraud accurately is almost an impossible mission.
It's no wonder MSN & Yahoo haven't set up a system like this yet - their waiting to see the outcome.
But with a combination of client side and server side monitoring, as we do and as Google could do, detecting a vast majority of CF is doable.
