homepage Welcome to WebmasterWorld Guest from 54.211.7.174
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Pubcon Platinum Sponsor 2014
Visit PubCon.com
Home / Forums Index / Google / Google AdSense
Forum Library, Charter, Moderators: incrediBILL & jatar k & martinibuster

Google AdSense Forum

This 32 message thread spans 2 pages: 32 ( [1] 2 > >     
Adsense FraudControl Team Ideas.
Was wondering why don't we formulate rules here.
aravindgp




msg:1468307
 7:46 pm on Jul 25, 2003 (gmt 0)

Hi Everybody

I was thinking long and hard about this.Why don't we formulate Fraud Control Rules here.Let's start with basic things like Average CTR should be so and so.Then may be we will all be very close to forming a good Rule sheet.I hope Googleguy would also make a formal opinion about this.

Ideas

1. Anything above 10% CTR is to be put on Alert.

2.

Expecting others to contribute.

Aravind

 

cornwall




msg:1468308
 7:51 pm on Jul 25, 2003 (gmt 0)

>>Anything above 10% CTR is to be put on Alert.

I don't think it is as cut and dried as that.

For a kick off it depends on the market sector your site is in.

Also it depends on the volume going through the site, very small volume may give high CTR.

In the end, as I have said on other threads, Google will have a very good fix now on what every industry secor can achieve on average CTR and on average CPC. Anything more than x% above that will be flagged.

Poweroid




msg:1468309
 7:58 pm on Jul 25, 2003 (gmt 0)

How about we formulate some "Alleged Fraud Response Rules" for Google? Any responses from GG most welcome ;-)

cramalot




msg:1468310
 8:01 pm on Jul 25, 2003 (gmt 0)

A percentage in not the way to measure.

For instance if a website get 1 visitor to a site in a day and that one visitor clicks on an ad the is a 100% click rate.

If a website gets 1000 visitors and one clicks, then that is a .1% click rate.

Both sites had the same ammount of clicks but the first one would get flagged? By what reason? For the most part and as a rule of thumb. The more the visitors - the percentage will tend to decrease. The margin will start to become greater.

My site has a high click rate. On other ads we have tested besides google. This isn't google ads.

On an article page with an ad can get from a 4% - a 56% click rate. We are wide spread but we do about 16.5% click rate for the magazine as a whole. We would get flagged. but we in no way spammed anything. It all comes down to targeting.

One

loanuniverse




msg:1468311
 8:01 pm on Jul 25, 2003 (gmt 0)

Ok, I will play:

2. Anything over two clicks from the same ip or 4 clicks from a range of ips within a 24 hour period will raise a flag.

cramalot




msg:1468312
 8:08 pm on Jul 25, 2003 (gmt 0)

I think click from IP is the way to go. The problem is:
A website that has many return visitors may tend to see this happen more.

So I think that in a 24 hour periond if a person clicks on an add on a specific site no more than 4 times. After 4 the clicks are not counted.

The owner's IP dosen't count towards a click.

Poweroid




msg:1468313
 8:09 pm on Jul 25, 2003 (gmt 0)

Anything over two clicks from the same ip or 4 clicks from a range of ips within a 24 hour period will raise a flag

Should raise a flag and discount further clicks, NOT bump the publisher. Can I play as well?

europeforvisitors




msg:1468314
 8:10 pm on Jul 25, 2003 (gmt 0)

I think fraud detection is one subject that Google's security people know more about than we do.

Our job as publishers is simply to publish and behave ourselves. :-)

cramalot




msg:1468315
 8:18 pm on Jul 25, 2003 (gmt 0)

I did and they are calling me a fraud.

As a God fearing person fraud is totally against my beliefs and morals.

Iguana




msg:1468316
 8:21 pm on Jul 25, 2003 (gmt 0)

I have been very impressed at the way Googleguy was able to tell one poster that his domain had been banned for using positioning software (and he seemed to know when it was used). If it's the same people who will investigate fraud then I'm okay with that.

IP sounds fine but AOL users have a restricted range of IPs for their cache and in fact when an AOL user clicks from one page to another the request comes from a different IP each time (just spent 2 days getting a web farm to move to cookie-based sessions from IP based sessions because of this). So lots of sharing of the same IP simultaneously. Also Freeserve UK have about a millions users and all are directed through a cache with limited IPs.

Poweroid




msg:1468317
 8:26 pm on Jul 25, 2003 (gmt 0)

I think fraud detection is one subject that Google's security people know more about than we do.
Our job as publishers is simply to publish and behave ourselves. :-)

That implies that webmasters who did get the letter are guilty of some fraudulent activity. I resent that.

Nobody doubts Google's security people's capabilities. What we regret is that they aren't being used.

loanuniverse




msg:1468318
 8:27 pm on Jul 25, 2003 (gmt 0)

I think fraud detection is one subject that Google's security people know more about than we do.
I agree, but you never know maybe there is a golden nugget there.... somewhere. Besides, everyone knows that this is just playing as Google makes its own rules.
Clark




msg:1468319
 8:58 pm on Jul 25, 2003 (gmt 0)

2. Anything over two clicks from the same ip or 4 clicks from a range of ips within a 24 hour period will raise a flag.

That's crazy I think. Some people like to click ads, some don't. If I go to a site on a topic and want to know what advertisers are in that field I will click every unique ad I can find. The Google security team will have a lot of work on their hands if they follow that rule. If the same IP clicked the same site 4 times I would discount 3 of them but not trip an alert.

I will play too, Not only you need to tie the CTR percentage to volume but you should also look at how many links out a page has. If there are a hundred links out and a CTR of 25% that is different than 10% on a page with NO OTHER outside links. So it would be worth breaking down automatically CTR on a per page basis. FR=FraudRank :)

europeforvisitors




msg:1468320
 9:29 pm on Jul 25, 2003 (gmt 0)

That implies that webmasters who did get the letter are guilty of some fraudulent activity. I resent that.

I think you need to learn the difference between "to infer" and "to imply." :-)

Visi




msg:1468321
 9:59 pm on Jul 25, 2003 (gmt 0)

Have been reading these posts with some interest, but starting to notice they are repetative and limited in participation.

Some comments about the basis of Google's interpetation. Since they are also serving the ads, don't you think they know the "expected" CPM? They have the basis for decision making from their own server stats and simple enough to compare. Tell me what all you want about the optimization of your site, but relative to their "search results" serving of the ads (sounds like the basis of adsense) they have well defined baselines. When you jump dramatically out of these ranges, they are going to look at why. All the other items are smoke and mirrors in relation to this fact. Bottom line is if you are better at serving targeted ads, than the group sering them to you, then some questions will arise. Just something to think about as the weekend approaches:)

cornwall




msg:1468322
 10:37 pm on Jul 25, 2003 (gmt 0)

<<Have been reading these posts with some interest, but starting to notice they are repetative and limited in participation.

I like that...mind you one could same the same about a lot of threads ;)

davewray




msg:1468323
 1:24 am on Jul 26, 2003 (gmt 0)

This is certainly a case for statistical analysis and I'm sure G uses various algorithms to determine fraudulent clicking. Say, for example, you have a small site and don't get many visitors/impressions per day. Over the course of a couple of weeks Google could use statistics such as standard deviations, means and normals for a certain smaller site. For a larger site, it would take fewer days to get a statistically relevant sample size to work from. If the site on a couple consecutive days goes outside, say, 3 standard deviations an internal flag goes up and they check out the situation closer. THEN, I say they should look at what ip's the clicks are coming from..AFTER statistical analysis..which, an algorithm can do on its own.

If a site gets flagged and Google discovers that a certain IP has been clicking on ads non-stop, it is VERY likely that the site owner has nothing to do with this. Google should place this IP on permanent ban so that if this person visits other sites with Adsense their clicks won't count either. This would curb the actions of those who just want to wreak havoc (perhaps your competition or some silly kid!).

If, however, Google determines that it is the site owner him/herself performing multiple fraudulent clicks...Good riddance and no mercy. Google has to play tough to keep Adsense alive and well...BUT play fair too! :) Just my few cents once again!

Dave.

europeforvisitors




msg:1468324
 1:46 am on Jul 26, 2003 (gmt 0)

If a site gets flagged and Google discovers that a certain IP has been clicking on ads non-stop, it is VERY likely that the site owner has nothing to do with this.

So one might think, but there are bound to be some technically clueless or just plain stupid people who have no idea that their clicks can be traced.

It's probably true, though, that the real problem with fraudulent clicks is likely to be more subtle. A friend just forwarded me a newsletter that describes a sophisticated clickbot that uses proxy servers to spoof IP numbers, spreads a "reasonable" number of clicks over a "reasonable" period of time, and supposedly can detect protection schemes used by advertisers. For a hundred bucks, a Web publisher can turn himself into a petty (or not so petty) criminal while minimizing the likelihood of getting caught.

I'm sure that Google has people working on algorithms to detect and defeat such programs, but fighting click fraud is probably a bit like fighting search-engine spam.

In addition to detecting and punishing fraud, Google might want to make it harder to commit fraud in the first place. For example:

1) Google could limit AdSense participation to sites that have been in Google's main index for a certain period (say, three months or even six months). This would make it harder to create disposable "click farm" sites, and it would make AdSense fraud less appealing to get-rich-quick types who haven't the patience to wait months for a payoff.

2) Google could require that publishers supply a tax identification number (e.g., a Social Security number) and the number of a major credit card at the time of enrollment. This would at least slow down small-time scam artists who try to open multiple accounts so they'll have an account to fall back on if their main account is cancelled.

cornwall




msg:1468325
 7:06 am on Jul 26, 2003 (gmt 0)

>>A friend just forwarded me a newsletter that describes a sophisticated clickbot that uses proxy servers to spoof IP numbers

This is indeed a big area. One that I was, in my innocence, not aware of till a couple of months ago.

A client was getting referrals from another site, the referral site was showing up big time on log analysis...

..but the referring site could never have sent that volume of referrals. I have the experience of looking at a lot of log files and a lot of referring sites...

..I looked at the raw log files and the info on IP numbers, and they were not all coming from the same place...

..I discussed it on one on the forums at WebmasterWorld, and a couple of people were kind enough to give me the details of the software one can use to spoof referrals.

..so step outside the bounds of what is reasonable, and they will probably ban you. They do not have to prove you are doing it, but outside their normal envelope, and perhaps a human look to see if there is noththing special about your site, and...Goodnight Vienna

Poweroid




msg:1468326
 12:34 pm on Jul 26, 2003 (gmt 0)

europeforvisitors, infer/imply - thank you for the English lesson.

My apologies that my English is not as good as yours. I was born and brought up in India where it's taught as a "second language".

Stop being pedantic. If the suggestion/implication/inferation - or whatever - remains that those who got the letter or got bumped are guilty, I still resent that.

richmondsteve




msg:1468327
 2:07 pm on Jul 26, 2003 (gmt 0)

...sophisticated clickbot that uses proxy servers to spoof IP numbers, spreads a "reasonable" number of clicks over a "reasonable" period of time, and supposedly can detect protection schemes used by advertisers...I'm sure that Google has people working on algorithms to detect and defeat such programs, but fighting click fraud is probably a bit like fighting search-engine spam.

I have no first-hand knowledge of Google's fraud detection algorythms, but I'm quite sure they have mechanisms in place to detect activity like that and they'll only improve over time. Let's not forget that while a site owner only has access to his/her logs, Google has access to statistics that span thousands of publisher sites. Patterns that aren't apparent when looking at a single site's activity may become obvious when looking at aggregate data.

richmondsteve




msg:1468328
 2:21 pm on Jul 26, 2003 (gmt 0)

A percentage in not the way to measure.

I agree. Even if CTR was a test to detect fraudulent activity Google would likely take into account other factors when determining whether a site warrants investigation. For example, they'd likely take into account trends in CTR for that site, the # of outbound links per page (less links would tend to mean greater CTR), range of CTRs for other sites showing the same ads, CTR by CPC range (to identify unusual click activity on higher payout ads), CTR by ad, etc.

Visit Thailand




msg:1468329
 3:42 pm on Jul 26, 2003 (gmt 0)

I am not so sure %'s are not a good way of looking at this. Of people I have spoken to who I know well, I asked them what they thought of the ads but they said they could not remember them and had not clicked any of them.

They had gone to the site for the content and had not really paid attention to the ads.

I asked a few business people what they thought and they really said the same as above.

For this reason I think that if you have a good content site which delivers content people are looking for then there will be a trend no matter what the industry in CTR.

richmondsteve




msg:1468330
 4:03 pm on Jul 26, 2003 (gmt 0)

I am not so sure %'s are not a good way of looking at this.

OK... I understand how your supporting points address effectiveness of text ads and users who ignore ads, but I do not see how the points address fraudulent activity, which is the focus of this thread.

For this reason I think that if you have a good content site which delivers content people are looking for then there will be a trend no matter what the industry in CTR.

Can you elaborate on what you mean by "there will be a trend"? Do you mean that everything else being equal text ads will become less effective over time and therefore CTRs will trend downward?

I'm still confused the connect to fraudulent activity, especially since you seem to be explaining a downward trend, while a downward trend would logically not flag an account for investigation for fraudulent activity.

europeforvisitors




msg:1468331
 4:20 pm on Jul 26, 2003 (gmt 0)

Stop being pedantic. If the suggestion/implication/inferation - or whatever - remains that those who got the letter or got bumped are guilty, I still resent that.

Nothing was implied. Those of us who didn't receive fraud letters from Google have no way of judging the innocence or guilt of those who did.

Poweroid




msg:1468332
 4:47 pm on Jul 28, 2003 (gmt 0)

europeforvisitors, I'm not looking for an argument and there's no way anyone can "prove" their innocence. Suggesting that keeping one's nose clean is sufficient to avoid Google's wrath is incorrect. We've seen here a variety of suggestions on how others' activities can get you banned. So even if nobody has proof that they are innocent or that they are victims of a third party activity it's agreed that someone else can get you thrown off. At least one webmaster has been reinstated suggesting that at least one of Google's flags turned out to be wrong.

You don't have to believe anyone's story about being innocent but not allowing webmasters here the benefit of the doubt is not helpful to anyone. Unless we accept the possibility that Google is sending these emails without any conscious violation by the publisher we won't discover the other possible reasons that they are doing this ...like certain javascripts or other technical reasons.

Taking a social security number/credit card number are good suggestions ..as is raising the admissions bar.

europeforvisitors




msg:1468333
 5:20 pm on Jul 28, 2003 (gmt 0)

You don't have to believe anyone's story about being innocent but not allowing webmasters here the benefit of the doubt is not helpful to anyone. Unless we accept the possibility that Google is sending these emails without any conscious violation by the publisher we won't discover the other possible reasons that they are doing this ...like certain javascripts or other technical reasons.

Is there anyone here who hasn't accepted the possibility that some of the Google e-mails were sent in error?

Accepting that possibility isn't the same as assuming that any given publisher is innocent or guilty. I have no way of making such a judgment, and neither do you.

In any case, as I pointed out in another thread, guilt (or innocence) may be irrelevant to whether an account is canceled for fraudulent clicking. If a site has repeated problems with fraudulent clicks from any source, Google may choose to cancel the publisher's account for two reasons:

1) To preserve advertiser confidence, and...

2) To minimize high-overhead (and therefore unprofitable) accounts.

Let's assume for a moment that someone is clicking on a site's ads because he has a grudge against the publisher or the advertiser. There's nothing the publisher can do to prevent this. So the best defense (really, the only defense) is to build a large, high-quality, comprehensive information site that earns significant revenues for Google. If a site looks professional, is clearly a great venue for AdSense, and is making serious money for Google, the scales are more likely to tip in the publisher's favor when Google weighs whether to keep or cancel the account.

richmondsteve




msg:1468334
 7:08 pm on Jul 28, 2003 (gmt 0)

If a site has repeated problems with fraudulent clicks from any source, Google may choose to cancel the publisher's account...

A very good point, europeforvisitors. It's no different than an insurance company who cancels your policy after you've made a lot of claims for a lot of money on your car. It may not be your fault that your car keeps getting stolen, damaged or hit, but the insurer doesn't care. They will drop you because of the perceived risk that they'll continue to lose money by keeping you as a customer.

We can all probably agree that ultimately Google's goal is to increase their profit. If the AdWords/Adsense program is more profitable when publisher sites associated with possible fraudulent activity are removed from the program that is their perogative, like it or not. There is a cost associated with investigating fraud, handling it and dealing with the affected advertisers and publishers and it might make business sense for them to do what europeforvisitors suggested for their own benefit and the benefit of the other advertisers and publishers.

As a publisher, would you prefer to have a $100 cut of the ad revenue generated via Adsense on your site and know there's a risk you might be booted for fraudulent activity caused by someone else or would you prefer an extensive and accurate fraud detection and management system and a $90 cut? What about $50?

Poweroid




msg:1468335
 7:24 pm on Jul 28, 2003 (gmt 0)

As a publisher, would you prefer to have a $100 cut of the ad revenue generated via Adsense on your site and know there's a risk you might be booted for fraudulent activity caused by someone else or would you prefer an extensive and accurate fraud detection and management system and a $90 cut? What about $50

That's going to be pretty size based. Big sites earning thousands a day from Adsense will go for the $100 as they are less likely to get chucked for fraud. Smaller webmasters like us earning less than $100 a day will probably opt to lose some revenue in favour of better fraud detection. I suppose some will argue that fraud detection should be part of the deal anyway; I'm not sure I disagree with them.

Commission King




msg:1468336
 1:35 am on Aug 4, 2003 (gmt 0)

Google doesn't want you to send targeted clicks. If they did, they wouldn't be booting the legitimate webmasters who know how to target their visitors well. Looks like google wants marginal traffic and perhaps fake or junk impressions to keep that clickthrough rate down. Not sure why they want the junk traffic, just appears that is what they want.

This 32 message thread spans 2 pages: 32 ( [1] 2 > >
Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google AdSense
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved