No no no, Scott. Click fraud is bad, and it needs to be fixed. And it can be.

But please... compared to magazine ads, radio, TV, etc., PPC is still a measurable environment. If you can't make money with it, you know right away... unlike when you place a magazine ad and wait 3 months to see if it worked.

Only the same stupid advertisers that have been paying for un-measuable media for decades are surprised and alarmed. And if they leave the PPC arena, we can get those click prices back down and aligned with realilty.

Maybe not good for Google, but the rest of us don't buy Google stock (too risky). :)

Brian, it's almost axiomatic at this point that PPC makes traditional offline advertising look like flushing money down the toilet -- at least the advantage of the old model is that you didn't know how much you were flushing, ignorance, bliss, etc. Now we got the dollars and cents measurements of waste, fraud, and abuse -- the sword cuts both ways thing.

Click fraud obviously needs to be fixed (especially for Google stockholders), but the question is -- can it be fixed? As long as the system is vulnerable to abuse, people will find a way to abuse it.

Click fraud was a revolutionary step forward, but why should the path end there?

Since I'm betting -- well, everything -- on local CPA advertising, we've thought through some of these issues. We think that CPA gets smaller advertisers at least out of the creative game. The publisher makes the best creative he can within certain parameters, as he should know best what sells on his site. Want to make your own creative? That costs extra.

That's not so far-fetched. ClearChannel has gone this route in local advertising in recent years, with some success.

And, CPA creates a near-perfect marketplace for ad inventory. What performs best runs most and in the best positions, and lesser-performing advertisers that want the juice pay an old-school premium for positioning.

Lotsa theory still, but we'll be practicing soon...

The debate over yield in CPA advertising has been raging for years in affiliate marketing, and it's been answered. The best convertising advertising, on an earnings per click or earnings per impression basis, gets the most inventory. Advertisers have to respond to this with good advertising, websites that convert well, and fair tracking, or publishers will give the space to someone else.

There's no doubt that you're asking the right big question, but perhaps we need to break it down a bit. There's two flavors of PPC -- search marketing and contextual advertising on third-part publisher sites.

Search advertising is great. Contextual advertising is not. Smart marketers have known from early on not to participate in the "publisher" side of Google AdWords (otherwise known as "AdSense"). It's bad ROI even without the fraud.

To make matters worse, that's where the fraud occurs -- on all the millions of junk AdSense pages. Google even teaches "respectable" publishers how to "blend" in the ads to prompt inadverdant and pure curiosity clicks, which, while not the type of organized fraud the report discusses, is still not in the advertiser's best interest.

PPC itself will come out of this fine -- in the SERPS. It's no surprise that Google is now rolling out CPA options, since even early-adopting AdSense publishers are moving back to affiliate marketing.

They know the AdSense party is almost over.

CPM &/or CPC = CPA for advertisers. In any case, if advertisers haven't CPA, they stop CPM or CPC campaigns ! For publishers, it’s the same; CPA must be transformed into CPC or eCPM commissions average.

One of the problems here is the definition of click fraud. Just because a click comes through a publisher that we consider less then "respectable" (another very subjective term), does not make it fraudulent. It may be less desireable in terms of conversion to a sale, but it is not fraudulent. Click fraud is something that Google and Yahoo are very aware of and spend time filtering out of the final billable click. So I am not too convinced by results of outshell's study. If I were an advertiser that did not have first hand knowledge into click fraud I too would likely say that spikes in clicks and a lower roi were the result of fraud.

The reality is that certain partners, delivery mediums, and placements will have different ROIs. The question is will Google/Yahoo want to give the advertiser the all the levers necessary to control these different factors.

That report is a load of crap.

http://plentyoffish.wordpress.com/2006/07/06/cl...

CPS - Cost per Sale!   It all comes down to the cost of advertising relative to the margins of the goods or services eventually sold. If the margin outstrips the CPS, then advertisiers will keep on doing it.

With Checkout, Google can make a very realistic guess at this.

So what?     Instead of offering the display ad to the highest bidder, per se, Google could offer it to the "most likely to make a profit" ... the advertiser who will keep on advertising at Google.

Expect Google to use that and change the game!

Michael Cooper (comment above this one) is right on.

The "creative will suck and the advetiser has no incentive to make it better" argument is dumb. CPA advertisers still have an incentive to create good creatives: they want to sell their stuff. And in a competitive bidding environment, with the ability to monitor conversions, (Google Checkout or integration w/ 3rd party services) won't google just serve the ad that converts to sales?

CT, you're right that click fraud is a "spectrum disorder" to borrow a medial phrase. Maybe we need a scale from 1 to 10, with 1 being a pretty crappy click and a 10 being a totally fraudulent bot click.

Markus, from the post on your site:

As both an advertiser and adsense user i’ve noticed a LARGE percentage of users will use your ad as a bookmark.

Do you have to pay for each of these “bookmark” clicks since they pass through Google? If you do, that’s insane.

Michael/Peter,

How can Google maximize efficiency if it doesn't control the middle piece, i.e. what happens on the advertiser's site after the click on the ad but before the purchase in Google Checkout?

Scott everyone does, in fact if you look up domain names of top brands you can see advertisers are bidding on their OWN domains.

At the end of the day NON of this MATTERS. Adwords is based on a market model and advertisers will pay the max price they can afford. If this percieved click fraud didn't exist then advertisers would bid 30% more and find something else to complain about.

Gosh, I feel baited :)

> Among the predictable responses

and

> it doesn’t matter (Danny) how big click fraud actually is — the system can’t escape the inexorable death spiral of negative advertiser perceptions.

Sure, perceptions are important. And they are key to understanding this report. About 400 advertisers were asked how much click fraud, on average, they felt they incurred. Overall, 15 percent was the average response. That's a big rate -- and used to come up with the $800 million figure, as I explained in my post that you referenced.

Now half of those advertisers also said they do nothing to analyze their clicks. So how did they come to the estimates they gave? They guessed.

That's like me wondering how much air escaped from a car's tires over a month. I ask 400 people to give me their estimates. 200 of them don't bother getting out the pressure gauge each month and actually checking. They just shout out whatever they think is right. And from that, you make some scientific conclusion on air loss?

That estimate is half-guess work and deserves debunking as such. As I said, for all we know, the situation is much less worse than described or equally much more worse.

But back to perceptions. Definitely interesting that about 30-40 percent say they'll cut some spending because of concerns. But the majority remain comfortable with it. Will that majority continue into the death spiral you talk about? Who knows.

What we do know is advertisers would like more options. CPA is coming not just as a solution to click fraud but because it's an option some advertisers simply would prefer to have. But what's the CPA you want to charge to a non-profit protest site, just trying to raise awareness of an issue through ads? They're after eyeballs, and that's why they might prefer CPM (and why Google gave them that in contextual ads some time ago).

No one ad pricing model is going to be the perfect solution for everyone. And there's always going to be some degree of fraud or inefficiency in everything. But today, the click fraud crisis has hardly faltered the industry, both in terms of spend or big legal payouts. I'm not saying it can't happen, but even despite click fraud (in whatever degree it is out there), search overall remains an extremely efficient advertising medium. That's why people are spending with it and pulling away from other areas to fund it (other areas that never, ever come under near as much examination about true circulation, wasted impressions and so on, because compared to search, you can hardly measure them at all).

Finally, the report had 63 percent at Google extremely satisified or somewhat satisified with the response they got on click fraud complaints. That's a pretty high figure. Google should make it 90 to 99 percent, of course. But that's not a figure that underlines a death spiral to me. And the other key figure, the fact that practically no one follows up with the search engines on complaints is important. The perception of click fraud is clearly strong. The actions advertisers take, according to this survey, is to sit back and do little except maybe reduce spending. More proactive moves on the part of the search engines might reverse that.

Danny, what a surprise, so glad you could join us :)

That’s like me wondering how much air escaped from a car’s tires over a month. I ask 400 people to give me their estimates. 200 of them don’t bother getting out the pressure gauge each month and actually checking. They just shout out whatever they think is right. And from that, you make some scientific conclusion on air loss?

If after a couple of months you've got a flat tire, the rate of air escape isn't really the issue. You know you've got a problem with your tires.

I’m not saying it can’t happen, but even despite click fraud (in whatever degree it is out there), search overall remains an extremely efficient advertising medium. That’s why people are spending with it and pulling away from other areas to fund it (other areas that never, ever come under near as much examination about true circulation, wasted impressions and so on, because compared to search, you can hardly measure them at all).

It will happen because new models will come along (soon) that will exceed PPC in efficiency the same way that PPC exceeded the old models in effciency.

I should make the distinction that you taught me, which is that "search" advertising is a seperate issue from PPC. I don't think annyone questions that a search results page a is a great place to deliver ads related to search terms. But how those ads function won't remain PPC forever.

Finally, the report had 63 percent at Google extremely satisified or somewhat satisified with the response they got on click fraud complaints. That’s a pretty high figure. Google should make it 90 to 99 percent, of course. But that’s not a figure that underlines a death spiral to me. And the other key figure, the fact that practically no one follows up with the search engines on complaints is important. The perception of click fraud is clearly strong. The actions advertisers take, according to this survey, is to sit back and do little except maybe reduce spending. More proactive moves on the part of the search engines might reverse that.

With all due respect, this really sounds like spin. Which is why I labeled your post predictable. This survey marks a point on a trend line -- what's your basis for concluding that the trend won't continue in a negative direction? "Death spiral" doesn't mean it's "dead" NOW, it means that things seem to be getting worse over time, so there is reason to expect that they will continue to get worse if the trend continues in the same direction.

CT, you’re right that click fraud is a “spectrum disorder” to borrow a medial phrase. Maybe we need a scale from 1 to 10, with 1 being a pretty crappy click and a 10 being a totally fraudulent bot click.

Here is the thing. The fraudulent bot clicks are largely if not completely fitlered out. In fact , if memory serves me correct, even a person clicking back and forth on the link repeatedly, does not cause additional billable clicks (on the old Overture anyway). Danny and Makrus are correct in stating that click fraud is largely a perception among advertisers.

The fact that 63% of advertisers are dissatisfied with the response on click fraud complaints is a tough one as the only answer that would satisfy them is to get some form of reimbursement. Makrus is also right in saying that even if the perceived click fraud did not exist the advertisers would find something else to complain about - and that is ultimately what is happening here. Unfortuantely, this is one of the cases where you have to trust the "black box" behind Yahoo and Google - which I realize is difficult if you don't have insight into it.

If after a couple of months you’ve got a flat tire, the rate of air escape isn’t really the issue. You know you’ve got a problem with your tires.

Sure, unless you aren't losing air at all, or enough air to notice. Any business is likely to have some degree of fraud, theft and waste. No one wants it, but the degree is what has the impact. We still don't know the degree of the click fraud problem.

Let me be clear. I'm not, absolutely am not, trying to downplay it as a serious issue. What I dislike, however, is being told there are some new industry stats that you can poke all types of holes in. This report gives me a figure that may be much lower than the real rate or much higher. Half the people not doing analysis but giving a click fraud rate is guesswork -- it's a stat based on guesswork. I want more than guesswork to quantify a problem. But that there is a problem, I don't disagree. We just don't know the degree.

But how those ads function won’t remain PPC forever.

I agree. In fact, I've written ages and ages ago about how at some point, you'll likely see flat rate purchases from those more after branding or those who don't want to play the bidding game. They'll be happy to write a big check just for the simplicity. PPC is not the only payment model, nor has it ever been (Google originally used CPM).

This survey marks a point on a trend line — what’s your basis for concluding that the trend won’t continue in a negative direction? NOW, it means that things seem to be getting worse over time, so there is reason to expect that they will continue to get worse if the trend continues in the same direction.

The point on the trend line is the percentage of advertisers who have said they actually have reduced spend. That also appears to be the first point, given this is the first time this survey has been done.

For all we know, if you asked this question to this group a year ago, even more might have said they reduced spend.

It is a concern, definitely, that so many people have said they reduced spend (though it would help to know more -- like was that spend on second tier engines? contextual ads?).

SEMPO, in contrast, has a survey that gives you more than one data point. Here's the write-up from 2004. For 2005, you have to be a SEMPO member to have the stats. But here's the summary from the slide deck:

+Nearly 2 out of 5 advertisers and nearly half of agencies have tracked fraud; 16% of advertisers say it's serious, nearly triple from 2004

+Yet a quarter do not believe it's a problem

+A third of advertisers and a third of agencies are concerned but not tracking fraud

Now that's your death spiral argument, at least that from 2004 to 2005, the percentage of those who thought it was a serious issue rose. But that might also be because awareness of the issue rose.

Stats also said that 40 percent of advertisers said they'd been a victim of click fraud (though note that a third of those surveyed said they weren't tracking at all). Stats also say that 80 percent have gotten some credit on complaints.

Overall, it's serious problem, one that alarms and concerns advertisers and that the search engines need to do more on, both in terms of preventing and also educating and changing perceptions (when half your advertiser just flat out guess they have click fraud, that's not a good thing). But I still come away from the stats with a mixed feeling of whether things are getting worse or not, in terms of the actual fraud happening. I guess we'll see if it's a death spiral or not.

The fact that 63% of advertisers are dissatisfied with the response on click fraud complaints is a tough one as the only answer that would satisfy them is to get some form of reimbursement.

It's the opposite -- 63 percent were satisfied with the Google response on click fraud complaints.

I'm coming late to the discussion, but I would like to point out that there is reason to be skeptical of the numbers that have been published so far estimating the amount of click fraud as a percentage of all clicks. There are quite a number of ways that click fraud can go undetected, because it is easy to manufacture clicks that just look like any other traffic that does not convert. This can be automated via spyware, or manually generated by a click ring.

My general opinion has always been that CPC is a poor business model. I am also surprised that Google did not realize that it would be so easy for fraudulent clicks to be generated, that they would be generated, that advertisers would become upset, and that lawsuits would be filed. It seems to me, as bright and talented as their engineering group is, they would have initially taken steps to at least provide advertisers both tools and options (such as CPA) to control their ad spend, thus reducing their exposure to fraud, or at least nonconverting clicks.