Palo Alto based Anchor Intelligence, which was called FraudWall until last week, may be part of the long term solution to the click fraud problem that plagues the big cost-per-click advertising networks (Google, Yahoo, Microsoft in particular).
Click fraud has been around as long as CPC marketing was first introduced by GoTo in 1998 (GoTo renamed to Overture and was acquired by Yahoo in 2003 for $1.7 billion). Google and Yahoo generally avoid the subject, but some outsiders estimate that as many as 20 - 25% of clicks are fraudulent - by people either trying to drive up the advertising costs of competitors, or by clicking on the ads displayed on sites they control.
Much of this fraud is detected by the ad networks and removed. But some of it isn’t. And it’s clear that the ad networks have conflicting incentives. Fraud directly drives revenue. Fixing it increases credibility, which is an indirect benefit. But the conflict naturally leads people to assume that the ad networks don’t fight click fraud as hard as they could.
Frankly, it’s not appropriate for the ad networks to be completely trusted to solve the problem, since they have a financial incentive not to. That’s where Anchor Intelligence comes in.
They, like competitors Click Forensics and Authenticlick, will be a neutral third party that audits the ad networks and report on fraud. Not only are they conflict free, but the fact that they will see traffic and clicks from multiple networks gives them an advantage. More data should mean better results.
Anchor Intelligence founder Ken Miller says they are working with fifty or so partner ad networks and advertisers. And he says a new big ad network partner will be announced shortly. Miller won’t say how they price deals.
The company has had two rounds of financing - a $2 million Series A in January 2007 from a slew of angel investors (Ron Conway, Baseline Ventures, Rajeev Motwani, Ram Shriram, Peggy Taylor, Jeff Jordan and Mike Maples). They raised an additional $4 million in September 2007 from JK&B Capital and earlier investors.
See all
Post
Never knew click fraud was such a problem.
I went to a search engine marketing seminar last month and the keynote mentioned that click fraud was considered to be quite low.
Obviously this is something that all advertisers should be aware of - with Googles attempts to reward based on Pay Per Action this could help stamp out a lot of the click fraud.
For the smaller advertisers you just know that some of your competitors will click on your ads to try and use up your monthly budget!
As sir jack welch said Integrity is required for long term success in personal life as well as corporate.
Think of it as with sarbanes oxley act U can get jailed for years altogether for comiting financial frauds with companies in responsible positions.
http://tekno-world.blogspot.com
“Much of this fraud is detected by the ad networks and removed. But some of it isn’t.”
Should actually read: “Much of this fraud isn’t detected by the ad networks and removed. But some of it is.”
Going to a CPA model will take eliminate CPC click fraud.
This is exactly what’s needed.
Hopefully more fraudsters will be caught while the innocent people who might normally be accused of click fraud are not.
Running a few simple queries on your log files will show if there are unusually large numbers of clicks coming from the same subnets or the same ISPs. Not 100% reliable, but is a good indicator.
“Fraud directly drives revenue. Fixing it increases credibility, which is an indirect benefit.”
As somebody building a search engine with a keyword sales interface I must disagree.
Fraud is not preferable. I am putting in place several algorithms not only to prevent fraud but to give users logs of click spam caught by the algorithms and removed from their SEO bill. Such as askimet does with word press.
This is only one of the many ways I am trying to make our search superior to Yahoo and Google, but it is an important one.
Click fraud is not good for anyone, be it the people selling SEO or those buying it and I doubt even Google would approve it in any embodiment.
What you wrote sounds like a conspiracy theory from an outsider trying to draw conclusions.
Mike, now this is the company you should be investing in.
ClickFacts is the clear leader in this space. We have more of the Internet’s largest advertisers as clients than anyone.
Others have more funding. But we have more revenue, and better systems for analyzing traffic and protecting the client’s brand.
There should be an alternative given to each advertiser to opt for a Pay Per Impression Model - and have those results equally integrated with the PPCs…
http://www.mattcutts.com/blog/.....ent-111708
Perhaps each advertiser could experiment with both for each ad - to analyze WHICH brings the best ROI.
Additionally search engines should offer information about the TIME SPENT on the Site after clicking on the ad - and WHERE they went to AFTERWARDS - to each advertiser.
Those clicks that show a stay of only a couple of seconds and NOT navigating to other pages after entering - should be ‘red flagged’ as suspicious
Ouuuuu! I hate auditors!
They come in and alwayy want to, to, to audit stuff!
btw
I’m introducing “Balm” on the site!
http://fakesteveballmer.blogspot.com
The real WTF is that they got a proper busniess model!
Ahhhh, this is the reason why I love this site, good post, no gimmick, direct to the point! Keep it up guys.
Nhick
http://www.itrush.com
I don’t know about other advertisers, but Google uses an auction system. The auction system should largely negate any click fraud problems. For example, if there were no click fraud, you might pay up to $1/click. If, let’s say, exactly half of your clicks are fraudulent, then you’d begin to bid just $.50 per click. Either way, you’d end up paying the same amount to acquire X new real customers. Anybody advertising on Google should be calculating their ROI and be willing to pay for an acceptable one.
The auction means that they cannot (or should not) pay more than they would want to for the real eyeballs.
Now, click fraud might matter if the fraud were unbalanced. For example, if your competitors were clicking on your ads. Then, their ads would be cheaper for the same keywords because theirs have less click fraud. On the other hand, it’s fairly easy for Google to catch this kind of click fraud if it’s major.
The other kind of click fraud you mentioned in the article, people clicking on ads on their own site, is perfectly unbiased. It is harder to catch, being many more people doing small amounts. However, it’s cancelled out by the auction. Every advertiser gets equally distributed click fraud. So, they each bid proportionally less.
Finally, click fraud is more of a problem if it changes a lot than if it’s just high. For example, if it’s 50% one week, 20% the next, and 35% the next, it becomes difficult to guess the value of the clicks in advance. Advertisers would constantly be overpaying/underpaying for clicks. Much more important than 0% click fraud is a steady (or slowly decreasing) click fraud. For example, if Google were to eliminate click fraud tomorrow (0%), then that would mean that instantaneously a lot of advertisers would be underpaying. Someone can swoop in, pay more for the more valuable clicks, and be over you on the ad sidebar. Of course, it shouldn’t take long for everyone else to increase their bids to reflect the higher value clicks.
If you think click fraud is a major problem, you don’t understand economics and statistics.
Of course, eliminating click fraud is useful to Google because it means that advertisers have to spend less time calculating the exact magnitude of real/valuable clicks. Trusting that close to 100% of clicks are real makes it much easier for small advertisers to quantify how much they would pay for one click. But, that’s the only thing that elminating click fraud does. It makes things easier to calculate, it does NOT mean increased profit.
As someone who used to spend $20K a month on AdWords (not Adsense), I have to say that click-fraud has never been an issue. The little weirdness that there was, Google corrected it.
All these click-fraud companies are praying on fear.
>As someone who used to spend $20K a month on AdWords (not Adsense), I have to say that click-fraud has never been an issue.
Really? And how did you come to that conclusion?
As for the statistic minded post before that, what he says would have more validity if ClickFraud were entirely balanced. It is clearly not true that “Every advertiser gets equally distributed click fraud”. ClickFraud varies widely by industry, advertiser, type of ad buy, and other factors. And ClickFacts’ focus is on the largest advertisers, those that suffer from the greatest amount of fraud.
Since different advertisers, publishers, keywords, and networks have different levels of fraud attached, by using ClickFacts’ services, you can take proactive actions to reduce fraud before it continues to occur, thus dramatically improving your company’s ROI.
Click-Fraud is a major problem as advertisers don’t have the tools to estimate how much of the clicks are fraud. Also, a third party, even when it’s used by many advertisers would still do worse in detecting click frauds since they only get information from the advertiser side, but they have no idea what the user did just before the click.
The people who claim that 25-30% of clicks are fraudulent have a direct stake in exaggerating the threat. In reality, these third party auditors have only a glimpse of the data that ad networks have and do a very poor job in differentiating clicks that are valid, invalid, or fraudulent.
I wish TechCrunch would practice a bit of due diligence and stop covering turds like this.
“”"
Since different advertisers, publishers, keywords, and networks have different levels of fraud attached, by using ClickFacts’ services, you can take proactive actions to reduce fraud before it continues to occur, thus dramatically improving your company’s ROI.
“”"
Snip.
I feel bad for you. You work at the company, so I know that there is no chance of convincing you. The cognitive dissonance is too great. I write the following for others who might otherwise pay you for snakeoil:
Of course different industry, keywords, etc. get different amounts of click fraud. But, for any one term, everyone bidding in the auction for that term is going to be bidding for the same value of a click. So, “mortgage” click fraud might be 50% because people who run sites where mortgage ads run get paid big bucks for clicking on their own ads. That means, every advertiser will pay 50% lower for that keyword. They might also run ads on the keyword “pottery”, which has just 1% click fraud. So, they’ll pay almost full value for every click.
The auction is Google’s brilliance. It literally means that you cannot overpay. (It also means that you cannot underpay!)
Back to ClickFacts. Instead of measuring clickfraud, which is low and unimportant, you should provide other metrics that will be useful for the clients. Specifically, you can gather data about click-throughs, amount of time staying on the site, etc, combine that with the estimated value of all these actions, and generate a recommended bid for the clients. Doing so correctly, and in a statistically coherent way, is not necessarily easy for most clients. The client wants to focus on her product, not auction theory and statistics.
CPA won’t fix the problem: lead generating advertisers will (and are already) plagued with bogus leads, see our article on botnets and why they generate bogus leads. Also, switching to the CPA model means moving the fraud downstream, as advertisers have incentive to minimize their reported counts (actions) to lower their bills.
Vincent, but as long as there is no collusion on a specific search term, then the auction system means that everyone will be paying exactly their max value for each real action even if they don’t report all the real actions (actually second price but it’s a small detail with a large number of players). Think about it. If I only report one in a thousand actions, then that means I only pay Google 1/1000 of how much I should. Everyone else who pays more, the real value, will be above me. I’ll get even fewer clicks, report even fewer. Until finally Google drops my ad entirely. That’s how Google’s Algorithm works.
Basically, simplifying for effect, you’re paying Google X dollars a month to be in position Y. If you pay less than X, then your position will be lower than Y or even dropped, which means that you get (more or less) proportionately fewer clicks/actions. If you try to “cheat”, you pay less, your position goes down, you get fewer actions, which means less revenue/profit. You pay less, you get less revenue. Then you just try to figure out how much you pay to maximize profit, not to minimize costs. Perfect. The auction ensures that you maximize utility not only for yourself, but for Google, and the economic systems!
Now, proof of no collusion: 1. It is difficult to collude considering the wide swath of people around the world who advertise on Google. 2. even if somehow somebody got every advertiser in the world colluding on a single message board, there would be a strong incentive to cheat since if you pay a penny more then you would get on the top of Google ads, then everyone else would cheat, and you’d get back to paying the max real value you’re willing to pay.
Economics should be mandatory. The great thing about auctions is that there is no flawed theories about mass social interaction to screw it up. It’s just: there’s money on the table, so someone will obviously take it. This leads to balanced fairness, it’s perfect.
Thank you for your input!
Clearly, ClickFraud is not low and unimportant, and we do provide all sorts of Advertiser Analytics in addition to ferreting out ClickFraud. ClickFraud was emphasized in this thread as that’s the context of the Techcrunch post. Nonetheless, if you look it strictly from an economic basis (and yes, I did major in Economics), ClickFraud generates money for the search engines and ad networks (and there are many in addition to Google), distributing it away from legitimate advertisers, and by crunching the data with our proprietary tests, we can clearly patterns of traffic which are not beneficial (and in some cases are fraudulent), which can successfully be worked around, thus increasing the advertisers’ ROI.
As you say, “There’s money on the table, so someone will obviously take it.” But the conclusion that folks taking money through ClickFraud is “balanced fairness, it’s perfect” only works if you’re on the side of the fraudsters.
When you go to the supermarket to buy raspberries, you expect 20% will be low quality, 10% will be garbage, in any basket that you buy. This is factored in in the price and nobody complains about it. Now let say that I have identified which stores offer the best quality, which days I should buy and at which time (just after a fresh delivery to the store), and how to pick out my basket (e.g. deeply burried baskets in the fridge are of better quality as the store try to sell the bad ones first). Now here’s what happens: I figured out a way to reduce my rate of bad raspberries from 20% down to 5%, without increasing my cost. At the same time, other buyers, not as sophisticated as me, are getting hit with rates of bad raspberries above 25%, because good baskets are bought by smart clients, leaving other buyers with below average quality.
This is exactly what is going on with bad clicks. As an advertiser, I hope I can pick up the good clicks cheaply and leave bad ones to less savvy competitors. For instance, I eliminate sources of bad traffic (sometimes difficult to detect, e.g. low frequency botnets) using Google levers. Indeed, my main competitor is spending 5x more money on PPC, and gets less than me (I get more for much cheaper in part because I’ve bought 50 times more keywords, efficient keywords in terms of return, but also I know better how to optimize campaigns and eliminate fraud).
“If you try to “cheat”, you pay less, your position goes down, you get fewer actions, which means less revenue/profit. You pay less, you get less revenue. ”
Wrong argument. If your break-even point is at 50 cents per click, it means you lose money when you pay 51 cents per click. However you make money when you pay 49 cents per click. Now if your profit per click is 0.5 cents per click when you pay 49 cents per click, it could be around 9.5 cents per click when you pay 40 cents per click, and around 39.5 cents per click when you pay 10 cents per click.
If you get 1000 clicks when you pay 49 cents/click, 500 clicks when you pay 40 cents per click, and 10 clicks when you pay 10 cents per click, your TOTAL profit is:
negative if above 50 cents per click
$0.00 at 50 cents per click
$5.00 at 49 cents per click
$47.50 at 40 cents per click
$3.95 at 10 cents per click
Even if you take into account that cheaper clicks have lower conversion rates, you’ll still see an optimum bid somewhere between 10 cents and 49 cents per click. And bidding above the optimum might increase your ROI but decrease your total revenue. And of course, bidding above the break-even point (50 cents in this example) will lead you to bankruptcy.
Smack my forehad! No auction in your calculations. You haven’t understood a single thing I said.
For simplicity, let’s assume the break-even of $.50 includes “normal profit”, that’s an economic term, look it up if you don’t know it: http://en.wikipedia.org/wiki/Profit .
If you bid less than $.50, say, $.40, then someone else, who sells the exact same thing as you do (this is the Internet, right?) will pay $.41 and place above you. In fact, someone will even pay more than him, and more than him…… until the auction goes up to your break-even of $.50. If you’re not willing to pay more than the $.40 (you should though since your break-even with normal profit is $.50), then you’re booted off of adwords since others are willing to pay that much.
Now, the real world doesn’t have such exact value metrics where we have infinite competitors with identical utility. Nevertheless, there are a lot, and they are spread out on a continuum. This ensures that the valuable top adwords space goes to those who value it most, and they pay Google as much as that space is worth (otherwise, someone will swoop in — remember that, subject to small second price differences of course). You can go look up more auction theory if you want to learn why.
It’s only cheating if there’s collusion. Otherwise, everyone wins.
Vincent @ Raspberries
You’re absolutely right. And if ClickFacts provides that service, as I said before, and as ClickFacts seems to confirm, then I think that is actually valuable to the clients. Yes, I agree that that is valuable.
The thing is, figuring out which keywords are best and at which times, etc, doesn’t really need to identify what is “click fraud” per se to be useful. For example, in analysis you might discover that clicks during 12pm EST don’t convert into sales (or w/e makes the real dough on your site). Or your analysis shows that clicks that you pay for for the keyword “platypus” don’t convert into sales either. It doesn’t really matter if the clicks are “click fraud” or if the clicks are simply legitimate users who don’t find your content interesting at that time of day/search term.
All that matters is, I’m paying $3/click, and 10 clicks/sale, $1/ sale, means costs > benefits, gotta change my strategy. This analysis is what you Should Be Doing(R) even if click fraud were 0%. I’ll repeat, even if Click Fraud were actually 0% you would still act exactly the same way. It’s the basic idea of, looking at how much actual real net profit you make as a ratio of clicks over all possible dimensions (time, keyword, website, etc.).
Now, my final point, and I’m done, is that this kind of analysis is not necessarily easy, especially with a large number of terms. Yes, the auction should ensure that nobody underpays, but the advertisers must accurately measure the utility of every click. That can be hard, as I said.
ClickFacts said above that they do this analysis for clients. That’s valuable. I’m glad that ClickFacts does this.
“Smack my forehad! No auction in your calculations. You haven’t understood a single thing I said.”
Aaarg. You have never run a business, have you? At 45 cents per clicks, my total revenue is too small and not worth the effort. If my competitors can make a good profit while I can’t, then good for them, they are smarter, but at that point the correct decision is likely to abandon the game. If they are stupid or purposedly running campaigns with negative ROI (one way to eliminate competitors if you have deep pockets), then I exit the game and try other, more effective marketing strategy. If they are stupidly burning money, I can just wait until they are gone, then I enter at a better price. A bit like people waiting 2008 to buy a new house. Makes sense?
mtcpgvois ytxm raslqg igrek lkehdo nleyvua hvaqwkr