30 replies to this topic

[Jill]

Andy Beal had an interview with Shuman Ghosemajumder from Google, who said that "invalid clicks at Google was 'on average is in the single digits, quarter over quarter.'

What do you guys think?

Here's the story:

Google's Invalid Clicks is Less than 2%

[Jbrookins]

Personally, I think he took the bait. Andy's a super smart guy, but it reads like the "exclusive" and "never before seen" data from Google caused him to leave his skepticism at the door. Later on in his comments, he even says "Google in effect is only telling us the numbers for click fraud they know about. There’s must be lots of fraudulent activity that goes unnoticed[sic] by them" which is something completely counter to his assertion of 2% (or "which means the actual number is more likely just a fraction of one percent!") as fact throughout the article.

From what I read in the article, he was shown some numbers/diagrams and told how they were derived. Maybe it's just that I'm a bit of a geek when it comes to data, but I find spreadsheets and empty explanations suspect. I am personally extremely skeptical of data that I or someone I implicitly trust did not produce.

Everything we have seen here points to Google considering very invalid clicks as valid.

In the real world, we have at least one partner that paid nearly $3,000 in a 48 hour period on traffic from sites that were blatantly against Google's TOS (domains unrepeatable on this forum). Adwords launched an investigation, told us "this does look awfully suspicious" and then denied that there was any fraud at all. This has happened more than once.

In the not-so-real world, we have actively committed click fraud (against ourselves) and got away with it. Google was oblivious then, when we weren't exactly sophisticated about it, so I have little faith in their ability to catch sophisticated click fraud.

Was the 30% we saw in our report (yeah, it was our report that these big numbers are cited from) an exception rather than the rule? Maybe. We were reporting a test that we conducted, which may have gotten blown out beyond our intent or it's actual validity with some repeaters. That doesn't mean that I'm gonna bite when Google says "See! Our data says 2%!" There's no way I would report Google's numbers as fact.

All in all, it's a great journalistic piece, but that's about it. At the end of the day, no one really cares about the clicks that Google catches as we (sorta) trust them to not charge us for those. It's the ones they DON'T catch that matter, and that article seems to try to minimize that issue by representing Google's 2% number as actual fact rather than claim.

[MaKa]

I must say I'm quite skeptical as well.

It's basically telling there is 2% click fraud which Google picks up after advertisers request a manual investigation. It doesn't say how much click fraud goes undetected - seeing they can't measure that because they don't know how much there is.

The process is a nice abstract of http://googleblog.bl...ilin_Report.pdf as far as I remember since I read the report a while ago. With the added information that the volume of manual investigation is about 2%, which Tuzhilin didn't mention.

Edited by MaKa, 12 December 2006 - 10:43 AM.

[jehochman]

In a properly managed PPC campaign, click fraud is a minor nuisance. It's just amazing how the dead tree media companies keep harping on this, as if they have no problems with circulation fraud. Yeah, right.

Whenever and wherever you advertise, some of the impressions are bogus. The savvy advertiser tracks conversion and only bids an economical cost per click or per impression based on the actual cost per conversion. If SEs allow too much click fraud, the advertisers simply lower their bids and the SEs make less money.

Edited by jehochman, 12 December 2006 - 11:52 AM.

[chrisbiber]

Jonathan,

AMEN to that. I think that the whole issue of Click Fraud / Invalid Clicks is hugely overhyped. For all of our clients we track conversions and cost per conversion, and optimize for both. In my experience, while we've had some invalid clicks, they have not negatively impacted the overall campaign performance.

Chris

[Jbrookins]

I agree that it's an over-hyped issue that isn't as doom & gloom as some people would like to think. We're not, on average, losing 30% of our advertising revenue to click fraud as some people have taken our particular report to mean. It's true that a large portion of click fraud occurs because the advertiser isn't particularly competent. HOWEVER, that does not change the fact that it does indeed occur and in much larger amounts than is being stated in this particular article.

Management competence is not at question here. This is about a claim by Google that only 2% of their clicks are invalid, and only a fraction of those are actually fraudulent. That particular assertion is laughable at best, and even if it was spot on, it represents a $200+ million a year problem.

I don't think this is a subject that can really be answered by "if you didn't suck, this wouldn't be a problem" and forgotten.

Edited by Jbrookins, 12 December 2006 - 12:25 PM.

[jehochman]

Competent advertisers know how to minimize click fraud. Meanwhile the search engines continue marketing these PPC programs to the public as if they are easily managed without expertise. That's the big beef. The search engines are profiting mightily from naive advertisers who overbid, choose the wrong keywords, and fail to set separate (lower) bids for the content network sites that can be hotbeds of fraud. The advertiser's loss is the search engine's gain. The SE has no incentive to warn people who are mismanaging their own accounts.

Edited by jehochman, 12 December 2006 - 02:53 PM.

[Jbrookins]

Aye, this is true to some extent, though it's a bit harder to avoid in large accounts in extremely competitive verticals.

The question still remains: Does the article deliver evidence of what it seems to assert, which is that there is only ~2% invalid clicks and a fraction of a percentage of fraudulent ones on Google? I'm voting for no.

[franco81]

I thought it was an excellently written article, sure it seemed like it was hyped as an inside exclusive with definitive answers about click fraud percentages straight from the horses mouth, but once I started reading it was easy to take it with a grain of salt.

Definitely agreeing with Jehochman about managing campaigns to minimise fraud. It seems like click fraud can affect certain campaigns more than others as it isn't much of a problem for us (running reasonably small campaigns) and that its a good idea to encourage people to get a professional to manage their ppc campaigns - especially if they are worried about the fraud thing.

I haven't heard mention of click fraud on other ppc platforms? Is it the same boat for Yahoo etc.?

[JACOBKELL]

Invalid or not,you can still be very easily banned by adsense by breaking any other rule.

[Betty]

Invalid or not,you can still be very easily banned by adsense by breaking any other rule.

well... if there wasn't a penalty of some sort for breaking rules then they wouldn't really be "rules" but more of guidelines.

[Martin C]

Google are becoming the masters of spin - this is in the same category as "There is no AIDs in Africa". In this case I don't think Google are entirely to blame because they have only not disagreed with someones observation of a Google produced chart that had no figures on it.

'Click fraud' is a stupid term - which is why I guess Google refer to it as 'invalid clicks' - but what is an invalid click?

I would say the problem with PPC is trying to eliminate, or give credit towards, those clicks where there is no intention or prospect of achieving a call to action. However a call to action might be a purchase, a service or even simply to promote brand awareness.

Charge per Action is not the answer.

Search triggered PPC is almost the perfect form of advertising but it has an Achilles heel.

Google benefit from invalid clicks more than anyone, they appear to have a minuscule team of engineers and investigators 'tackling' the problem that they also don't see as a problem and they are quick to take a shekel for any Tom, Dick and Harry Made for Adsense website.

For Google to effectively eliminate the problem, which I believe they could do, would mean a loss of revenue.

Everyone has a choice to advertise with them or not, but Google should have responsibilities too and I think advertisers would be foolish to accept this headline figure. Google are not alone at not tackling the 'invalid click' problem but they are the company that benefits the most.

Advertisers are being ripped off but most advertisers are going to be unaware and therefore not report any queries, if they are not complaining Google are not going to start looking. Advertisers can do things to minimise their exposure but many people are advertising not understanding the problem and simply accepting the SE view that it is a problem that is under control.

I suspect like the tobacco industry denying the link between smoking and cancer, we will in time eventually find out they knew all along, Don't be Evil? - give me a break! The sad thing I find is that a company that is bringing in over $3,000,000 a day, every day - can't do more than play the 'spin' game.

[projectphp]

I think Google are pretty right myself. Click fraud is not a widespread mass phenomenon. It simply isn't. I have seen no evidence that any generalised figure, say 8% click fraud, exists accross the board for every account, and I would wager that the figure of 2% is even high for most accounts.

What click fraud is is an isolated phenomenon, in which some accounts get hammerred, and many accounts receieve little or none.

Advertisers are being ripped off

Which advertisers? Certainly none of my clients, who receive excellent ROI from AdWords. I would say, if anything, they usuallly could afford to pay Google more, and probably would if Google chose an affiliate model. Besides which, if the click fraud was killed, they would just spend more per click.

many people are advertising not understanding the problem

And whose fault is that? No one holds a gun to people's heads and makes them use AdWords, or the much more click fraud prone AdSense. If people choose to advertise with Google, they accept the risk, of which ignorance/lack of skill, not click fraud, is the biggest.

I would say the problem with PPC is trying to eliminate, or give credit towards, those clicks where there is no intention or prospect of achieving a call to action. However a call to action might be a purchase, a service or even simply to promote brand awareness.

Martin, with all due respect, that is the job of the advertiser. PPC is already dumb enough, and if it moves away from requiring any skill, everyone loses.

Google have, IMHO, the ultimate system: one that is infinitely easy to use, yet infinitely complex to master. Any move to dumb it down would hurt more businesses than it helps.

Where your argument above falls down is that many people are bad at AdWords independant of click fraud. I had a client that sold in Sydney exclusively. I put in -melbourne and other Australian cities, and overnight he dropped significant levels of (non-converting) traffic, and increased lead generation. That isn't the medium's fault, that is the advertiser's.

Similarly, conversion rates play a part. Two sites, in exactly the same space, can have vastly different conversion rates for exactly the same term. And that is the beauty of PPC. One is working in a system where ROI can be improved independant of Google, with improvements to the actual site leading to better ROI. Unlike CPA (Cost per acquisition), improvements in conversions help improve the ROI of PPC, as more traffic converts from the same volume of clicks, where as CPA benefits from any conversion improvements linearly, that is a 1% increase in conversions is a 1% increase in payouts. Smart businesses can therefore often get a better return from PPC than they could from CPA.

IMHO, click fraud is like shoplifting: no one likes it, but everyone needs to accept that a small percentage is inevitable. After all, I could click on an ad right now for no reason but to show one example of undetectable click fraud. I mean, how would you stop that?

<added>http://www.traffick....s-exclusive.asp Andrew Goodman's take, which is very similar to my own, right down to where most comments come from, and the view that third party click fraud is (more than likely) a watse of time. Well worth a read.</added>

Edited by projectphp, 14 December 2006 - 08:07 PM.
Added Andrew's Take

[oneofthe3lions]

It is possible that they simply believe that 2% of webmasters are 'dodgy', therefore equating to 2% of everything is bad, click fraud, spammy websites, blackhat redirects , pop up sites etc etc. Not very scientific but I could get my head around that.

[projectphp]

If you read the article in question (always hard, I know), it shows that the overlap between invalid clicks and click fraud is almost totally a match. The 2% is the bit they miss (much lower in reality is the claim).

Heck, allow me to show you the image:



Google also discuss fraud misnomers (like multiple IP address clicks and page reloads), and this is fascinating, at least to me, because I think most click fraud detection is, at best, dodgy and, at worst, exactly what they accuse Google of: having secretive, unjustifiable formulae.