homepage Welcome to WebmasterWorld Guest from 54.243.12.156
register, free tools, login, search, subscribe, help, library, announcements, recent posts, open posts,
Subscribe and Support WebmasterWorld
Home / Forums Index / Google / Google AdWords
Forum Library, Charter, Moderators: buckworks & eWhisper & skibum

Google AdWords Forum

This 37 message thread spans 2 pages: 37 ( [1] 2 > >     
Click Fraud - A Billion Dollars a Year in Invalid Clicks
10% Of All Clicks Identified as Invalid - Advertisers ID less than .02%!
Brett_Tabke




msg:3267418
 4:55 am on Mar 1, 2007 (gmt 0)

[adwords.blogspot.com...]

Our invalid clicks rate ; the activity rate ; has remained in the range of less than 10% of all clicks every quarter since we launched AdWords in 2002. At Google’s current revenue rate, every percentage point of invalid clicks we throw out represents over $100 million/year in potential revenue foregone.

Because it is difficult to definitively determine the intent of a click in many cases, the number of invalid clicks that we filter also include those filtered for reasons separate from fraudulent intent. Cases of provable click fraud attempts constitute a small minority of the clicks we mark as invalid. There are many greyer cases of possible click fraud attempts (but without clear scientific proof, for which we still choose not to charge advertisers. For example, we have an automated rule which filters out the second click of all double clicks as a matter of policy. We mark this kind of activity as invalid simply to optimize advertiser ROI. Those clicks are included in our activity metric and are also a good reason we use the term invalid clicks instead of fraud.

MarketWatch [marketwatch.com]
Reuters [news.yahoo.com]

But according to Google, in the worst cases, on average 10% of all ad clicks are invalid. Typically, the amount is in the low-single-digit percentages. Google bases that figure on the average number of invalid clicks that it catches, and as a result, doesn't charge customers for. That amounts to about $1 billion a year in payments Google could have collected, but chose not to, it said.
The amount of click fraud Google doesn't catch, but is brought to its attention by advertisers, represents less than 0.02% of the times an ad's clicked upon, according to Google.

Google also said Thursday it will introduce a number of new click-fraud-fighting measures in coming weeks and months. They include letting advertisers notify Google about specific Internet addresses from where they suspect click-fraud attacks are emanating.

Cnet [news.com.com]

Google claims that click fraud represents a very small amount--a percentage that is in the single digits--of total clicks, and says it catches nearly all of it before customers get charged. Google has said that less than 10 percent of all clicks on ads it serves are dubious in nature and it does not charge advertisers for those. Google provides refunds to customers who request them because of suspicious clicks for less than 0.02 percent of all clicks, the company said.

Forbes [forbes.com]

On Wednesday the company said that "less than 10%" of all clicks in its AdWords network are “invalid". What's an invalid click? Not necessarily a fraudulent click: It could also be unintended double-clicks, certain "innocent" types of bots or web-crawlers that open up advertising links, or simply Web browser errors.

OK, so how much click fraud is Google dealing with? The company still won't say. It's a subset of that "less than 10%" number, but it won't say more than that, other than noting that advertisers themselves identify .02% of their clicks as fake. If Google agrees an advertiser-reported click is truly bad, it offers a credit.


 

walkman




msg:3267426
 5:03 am on Mar 1, 2007 (gmt 0)

>> but is brought to its attention by advertisers

totally misleading. How many advertisers can analyze (or choose to) the data and take the issue up to google?

Brett_Tabke




msg:3267427
 5:07 am on Mar 1, 2007 (gmt 0)

Exactly:

The percentage of invalid clicks actually identified by customers is 0.02 percent of all clicks

So they are saying 10% are invalid. I'd say they are admitting that Click fraud is 10%.

kdobson99




msg:3267428
 5:09 am on Mar 1, 2007 (gmt 0)

Almost as big of an issue is distribution fraud - Google not controlling where your ad is appearing in a proper manner. Just look at all of the arbi sites running Search Partner feeds.

For a really cool demonstration video, go to youtube and search for "ppcabuse" (its not mine, but exactly what I am talking about)

blend27




msg:3267450
 5:58 am on Mar 1, 2007 (gmt 0)

So now they want to know(just to make sure) that i know what is wrong with them?

BTW: The IP address that i spent most of the time translates to "localhost" and I speak 6 languages(human) with the ability to spot the MFA (nothing against MFASHOP :) ) within 3 seconds to landing.
I could be as government, WIFI and Major ISP(s).

SO.

I search for scrapers that were reported and BANNED 4 MONTH AGO, STILL ADS by ...

'G' 'Y' 'M' - FLEX YOUR MASSES....

rocknbil




msg:3267499
 7:49 am on Mar 1, 2007 (gmt 0)

Well this newbie will add this.

Keep your ads off the content networks. Everyone is trying to make money off adSense. It seems like we get a lot of reliable valid traffic just on SERPS ads. There is nothing to be gained by clicking on a Google serp ad for the fraudsters, other than trying to bring us down by attacking our budget.

Real world case of "oops my bad" - I set up adSense for a former employer. When he expressed concerns, "how do I know if it's a competitor?" I explained the methods for determining this. I then implored, vehemently, don't click your ads. Use your adSense login. Don't click your ads. And once again, if you are thinking of clicking your ads, don't.

Two days folks. They were banned in two days because he just couldn't resist, it seemed so much easier to just ease his curiosity by clicking his ads than to log into his account and use the penalty-free way. So Google does try, but as usual - no barriers for the truly motivated.

koan




msg:3267506
 7:58 am on Mar 1, 2007 (gmt 0)

Brett you seem to have put your own spin on the news title. Those 10% invalid clicks are identified as such by Google and are not charged to advertisers. What they are saying is that maybe 0.02% of those clicks are invalid *and* charged to advertisers. That sounds like peanuts and as long as advertisers have a decent ROI, I don't see what's the big deal, as long as Google is diligent with banning fraudsters. No advertising medium is 100% efficient.

benflux




msg:3267507
 8:01 am on Mar 1, 2007 (gmt 0)

The ads in the video are not Google surely? Too long....must be Yahoo?

Moncao




msg:3267509
 8:10 am on Mar 1, 2007 (gmt 0)

This used be an seo method on Yahoo and a way to get at your competitors on ppc. PC, bespoke click software, anon proxy list, revolving browser type, click once on unrelated search to get cookie, retain cookie / browser type / proxy and click again for near related search and to upgrade cookie to next level, third click is the target click (in Yahoo's case you would then appear to stay on that site for ever as your cookie / IP / browser type never went back to Yahoo).

trinorthlighting




msg:3267722
 1:27 pm on Mar 1, 2007 (gmt 0)

At least google is divulging some information about click fraud.

Brett_Tabke




msg:3267745
 1:39 pm on Mar 1, 2007 (gmt 0)

> What they are saying is that maybe 0.02% of those clicks are invalid *and* charged to advertisers.

Not the way I read it - nor the way Market Watch or Reuters do either. They are saying advertisers only ask for .02% back as bad clicks. That's not a click fraud rate - that is just the rate that advertisers find it in cases where google doesn't. Google says they identify 10% as bogus clicks in one form or another. They are admitting that click fraud is potentially 10%, but advertisers are so blind to the situation, that they only ask for .02% back.

It is a fine line for Google to walk. They are making the case that it is only fraud if they don't catch it and the advertiser reports it. eg: that they find 98.8% of it an only .02% is found by the advertiser.

What this article does is send a signal to the investment community that Googles exposure to future liablity is minimal.

wruppert




msg:3267748
 1:41 pm on Mar 1, 2007 (gmt 0)

Soley an advertiser, I dropped out of the content network ages ago. More recently, I also dropped out of the "Search Network" as well. It seemed that many of the search partners were putting up just page of ads with minimal organic results.

That ppcabuse video mentioned above was very interesting, and well done.

Receptional




msg:3267797
 2:27 pm on Mar 1, 2007 (gmt 0)

I'd say they are admitting that Click fraud is 10%.

Gotta disagree with you there Brett. They are saying 10% of clicks are invalid or not chargeable. Not necessarily refundable because most are filtered at source and never charged anyway.

When I get stung on PPC fraud it is a) rare and b) painful. But mostly it is a) rare and b) insignificant.

Most "PPC fraud" is actully dumb PPC managers bidding on broad match.

[added]we don't use content match much, I'm afraid... so I can't comment on that, except that it sucks for conversions[/added]

[edited by: Receptional at 2:31 pm (utc) on Mar. 1, 2007]

Brett_Tabke




msg:3267803
 2:32 pm on Mar 1, 2007 (gmt 0)

> Not necessarily refundable because most are filtered at source and never charged anyway.

So if a bank rober is stopped before he walks out of the bank with money - isn't not bank robery and the whole thing is forgotten? Just because they aren't charged, doesn't remove the fraud classification.

It is a fine line for Google to walk. They are making the case that it is only fraud if they don't catch it and the advertiser reports it. They are invalidating that 10% of a reason - mostly fraud. eg: that they find 98.8% of it an only .02% is found by the advertiser.
Just because they find 10% doesn't mean it isn't attempted click fraud.

ByronM




msg:3267810
 2:37 pm on Mar 1, 2007 (gmt 0)

MSN and Yahoo can really push back Google if they were to split out there content networks from there publisher networks and allow people to bid to play on premium content owned by either of them.

I belive MSN is getting close with the beta adcenter, just don't know when that goes live. I'm sure that could drive down click fraud since there would be no gain to the fraudster (unless your simply abusing the clicks to financially break your competitors of which it would be easy to detect for the most part in automated fraud sweeps/trending/analysis)

Receptional




msg:3267822
 2:50 pm on Mar 1, 2007 (gmt 0)

They are invalidating that 10% of a reason - mostly fraud.

Again - I don't see them saying that. Back clicking is an incredibly natural user activity, for example... they see results, click on a result, view it for a second, back, click on another, view it... (paid and natural) and then click again on the original paid link.

That's not fraud. It's invalid clicks.

99% of search users don't have a clue what the differnce is between the two types of click. But I know this much - even if I am not first in the organic results, I'll get traffic. Why? because the user doesn't go to the first result? no - because she back clicks, to quickly decide which site is right for her. She'll end up double clicking on organic results at a very similar rate to paid, I bet.

walkman




msg:3268033
 5:46 pm on Mar 1, 2007 (gmt 0)

>> They are admitting that click fraud is potentially 10%, but advertisers are so blind to the situation, that they only ask for .02% back.

Brett,
attempted fraud (or fraud to Goog) is 10%--no one is charged for that, google loses it. On top of that, 0.02% ask for more, meaning fraud that escaped Google. I believe that many mom and pop stores don't know how to or see it as cost of doing business, making the 0.02 much, much higher IMO.

The best way to calculate fraud is connect it to the % of spam pages that rank high on Google :) That's the same rate of fraud I'd bet.

TypicalSurfer




msg:3268055
 6:11 pm on Mar 1, 2007 (gmt 0)

I sense hoof beats chasing the fraud story at google, I doubt they would go pro active like this if something wasn't on the horizon.

This is amusing:

computers automatically detect virtually all such problems, meaning advertisers pay nothing.

I think most of us know better than that.

SEOVisits




msg:3268064
 6:16 pm on Mar 1, 2007 (gmt 0)

attempted fraud (or fraud to Goog) is 10%--no one is charged for that, google loses it. On top of that, 0.02% ask for more, meaning fraud that escaped Google. I believe that many mom and pop stores don't know how to or see it as cost of doing business, making the 0.02 much, much higher IMO.

Exactly, most mom & pops just have to settle with what they get from Google & others as there is no way they can refute it. They don't have the data, expertise, or tools to figure out they are getting robbed. IMO, there needs to be standards set for the search engines & it is only a matter of time before it happens.

europeforvisitors




msg:3268277
 9:35 pm on Mar 1, 2007 (gmt 0)

So they are saying 10% are invalid. I'd say they are admitting that Click fraud is 10%.

They aren't admitting that all. They're saying that, on average, somewhat less than 10% of clicks are invalid for any number of reasons, with click fraud being merely one of those reasons.

A simple (and hypothetical) example of a "grey area" click incident might be two clicks in rapid succession, which could occur if an impatient user clicked an ad twice while waiting for a response. The second click might be "invalid," but it wouldn't be fraudulent.

mimmo




msg:3268428
 11:09 pm on Mar 1, 2007 (gmt 0)

So they are saying 10% are invalid. I'd say they are admitting that Click fraud is 10%.

Maybe I missed the point here, but 10% of what?
Does the total include also the Content Network clicks?

On "Google Search" related ads, who has interest on clicking on the ads, beside Google itself or your competitors? I think fraud here is not very high.

On the Content/Search Network it is a different story.

I would not be surprised if "invalid" clicks are something like 30% on the Content Network and 5% on the Google searches.
How about stats by country?

Maybe it is just me ... but I am more concerned about the value of the clicks you are paying for, which Google things are "valid clicks". If they are not converting, they are 100% useless even if "valid clicks".

Dennis Reinhardt




msg:3268458
 11:47 pm on Mar 1, 2007 (gmt 0)

I read the statement:

Google provides refunds ... for less than 0.02 percent

differently than some above. There are two conditions here: (1) the customer requests a refund *and* (2) Google actually pays up. My experience with Google "customer service" is they ignore evidence submitted (log files) and go with their own internal findings.

There is a number missing here: what percentage of transactions are requested by customers but denied by Google?

FourDegreez




msg:3268539
 1:02 am on Mar 2, 2007 (gmt 0)

A simple (and hypothetical) example of a "grey area" click incident might be two clicks in rapid succession, which could occur if an impatient user clicked an ad twice while waiting for a response. The second click might be "invalid," but it wouldn't be fraudulent.

I would imagine this is relatively common, as may be other common behaviors that Google may not "count". To paint all such clicks as "fraud" is a gross misrepresentation. From the quoted article:

Cases of provable click fraud attempts constitute a small minority of the clicks we mark as invalid.

And furthermore, probable or possible click fraud attempts consistute another portion of the 10%. The rest of the 10% is rounded out by garden-variety junk clicks. In any case, none of these clicks are charged to advertisers, so I can't see how anyone would read this and take it as evidence against using the content network. It looks to me like Google is over-cautious in not charging on clicks if there is any reason to suspect those clicks aren't legit.

lexipixel




msg:3268781
 9:06 am on Mar 2, 2007 (gmt 0)

My local newspaper, which is owned by a larger parent company, (which owns many, many other newspapers), has a "Print and Deliver" program. They print 8 1/2" x 11 glossy fliers and insert them into the paper.

I can tell you that if there are 10 of them in the papers for one week, (one or two some days none, Sunday a few).. that at least 2-3 of them are "doubles".

Yup. The print newspaper promises to print and deliver 1,000 or 10,000 or 100,000 flyers, and a good percent, (let's say... "less than 10%"), of them get stuffed 2-3 at a time into (1) newspaper.

This is about the same thing.

Magazines use the term "total distribution" to include every free copy they dump on the street -- this is a different metric than "total paid subscriptions"... even if it is only "less than 10%".

Again, much like "invalid clicks".

My $0.02:

(And isn't it funny that the amount that Googel says advertisers reported was 0.02?)...

...anyway,

An invalid click is not necessarily click fraud, but a click fraud click is an invalid click.

(Say that three times fast).

gibbergibber




msg:3269000
 2:24 pm on Mar 2, 2007 (gmt 0)

--So if a bank rober is stopped before he walks out of the bank with money - isn't not bank robery and the whole thing is forgotten? --

Interestingly, this is exactly the case for shoplifting.

Store detectives usually wait until someone actually goes past the checkout with a stolen item until arresting them, because that's when the crime has actually taken place.

mitchy231




msg:3269131
 4:14 pm on Mar 2, 2007 (gmt 0)

--So if a bank robber is stopped before he walks out of the bank with money - isn't not bank robbery and the whole thing is forgotten? --

I see the correlation, however in the analogy used above, wouldn't it be the robber stopping himself (Google saying they identify 10% of fraud)?

In that case no crime would be committed. I do see Brett's point. They are setting themselves up to rebut future legal action. The "well we told you" statement.

Either way, my ROI has been great over the last 6 or so months. However, I do keep a very close eye on fraudulent activity.

gregbo




msg:3269804
 7:53 am on Mar 3, 2007 (gmt 0)

Not the way I read it - nor the way Market Watch or Reuters do either. They are saying advertisers only ask for .02% back as bad clicks. That's not a click fraud rate - that is just the rate that advertisers find it in cases where google doesn't. Google says they identify 10% as bogus clicks in one form or another. They are admitting that click fraud is potentially 10%, but advertisers are so blind to the situation, that they only ask for .02% back.

My reading of the post on the AdWords blog is that Google has identified 10% of all clicks to be invalid (by their definition) and thus, they do not charge for them. 0.02% of all clicks are the result of complaints by advertisers of click fraud; manual investigations by the click quality team reveal that they should be included in the invalid category and refunded.

This, of course, does not answer the question of the clicks that Google did not identify as fraudulent, but were fraudulent, either because no one complained, or their filtration/click quality did not identify them as invalid. Such clicks can be manufactured quite easily by botnets operated over a large distribution of IP addresses, user agents, language types, etc. They can also be generated by humans operating under similar instructions as the botnets.

gregbo




msg:3269810
 8:04 am on Mar 3, 2007 (gmt 0)

And furthermore, probable or possible click fraud attempts consistute another portion of the 10%. The rest of the 10% is rounded out by garden-variety junk clicks. In any case, none of these clicks are charged to advertisers, so I can't see how anyone would read this and take it as evidence against using the content network. It looks to me like Google is over-cautious in not charging on clicks if there is any reason to suspect those clicks aren't legit.

Here is an analogy. If someone owned a car, and 99.8% of the time it ran fine, but 0.2% of the time the car not only failed to work, but caused the individual some type of financial loss, said person might very well get rid of that car and buy a new one. Word might spread and others wouldn't buy those cars either. Such is the case with the content network. It has been tainted because it was not always the case that G was so diligent about tracking "invalid" clicks (if they are even able to do that as well as they say).

mimmo




msg:3269907
 11:35 am on Mar 3, 2007 (gmt 0)

Such clicks can be manufactured quite easily by botnets operated over a large distribution of IP addresses, user agents, language types, etc. They can also be generated by humans operating under similar instructions as the botnets.

Yes, I guess it is not that difficult to generate clicks which are not detectable as fraud: different IP, no cookies, no javascript.

But this type of fraud would only apply to the content network wouldn't it?

If Google will finally allow advertisers to select and limit which content network websites you want your ads to run on, advertisers would have the ability to stop those websites that do not generate revenues.

Two websites, both convert at 10%. I do not care if the other 90% is fraud, invalid clicks or uninterested customers. What I care is the 10% conversion rate. To the extreme, if a website brings me 10 customers and artificially creates 90 extra clicks, I am fine with that as far as my ROI for that website is what I want.
I guess I am asking for as much control and transparency as possible, so I can take my own decisions.
Right now I do not think the content network is transparent and controllable enough.

europeforvisitors




msg:3270052
 3:50 pm on Mar 3, 2007 (gmt 0)

But this type of fraud would only apply to the content network wouldn't it?

Not according to Google's CFO. From a CNNMoney.com article [money.cnn.com] on click fraud (relevant phrase highlighted in bold):

The main perpetrators appear to be competitors of advertisers and also scam sites set up for the sole purpose of hosting ad links provided by Google, through its AdSense unit, or Yahoo!, through its Overture service.

If anything, "competitor fraud" is probably easier to do on SERPs than on the content network, because the perpetrator knows the domains where the ads will appear (Google.com, Google.co.uk, or whatever) and which keyphrases will trigger the ads.

If Google will finally allow advertisers to select and limit which content network websites you want your ads to run on, advertisers would have the ability to stop those websites that do not generate revenues.

That's coming soon, according to this AdWords Forum thread. [webmasterworld.com]

[edited by: trillianjedi at 1:56 pm (utc) on Mar. 6, 2007]
[edit reason] Fixing formatting... [/edit]

This 37 message thread spans 2 pages: 37 ( [1] 2 > >
Global Options:
 top home search open messages active posts  
 

Home / Forums Index / Google / Google AdWords
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About
© Webmaster World 1996-2014 all rights reserved