On Monday, ComputerWorld reported that Google is partnering up with click-fraud detection firm Click Forensics in order to aid in the detection and reporting of search ad click fraud.
Also, Google has stated that it will now accept submissions from Factr, a product of Click Forensics that was created with the help of Yahoo, that gathers and submits click-quality reports. Through this reporting process, Google is hoping to help advertisers receive refunds for click frauds.
As a side note, Yahoo has previously paired with Click Forensics back in March, and was the first engine to allow a 3rd party to obtain additional feedback for the purposes of improving advertiser satisfaction and ROI
Recent controversy has been swirling over the real age of China’s Gold-Medal gymnast He Kexin, who edged out U.S. gymnast Nastia Liuken for the gold medal in the un-even bars.
The youngest age for women to compete in the Olympics is 16-years-old, but several Internet sites, news reports, and even former U.S. womens gymnastics coach Bela Karoli are alleging that Kexin was born in 1994, making her 14-years-old and ineligible to compete in the Beijing games.
Although Kexin’s passport indicates that she was born in 1992, speculations about the girl’s real age have fueled an astonishing investigation on world-wide search engines.
A blogger known as ‘Stryde Hax,’ began an independent probe into the ages of the Chinese gymnasts by searching for primary state-issued documents. His claim is that once these official documents were made public, he should have been able to easily search and access them on the Internet.
Stryde first searched Google.com, asking it to search all Chinese web pages for “He Kexin” and “1994”. He reported that Google returned one result, an Excel spreadsheet listing all of the Chinese Athletes’ birthdates. Upon clicking on the result, he discovered that the document was no longer available, but he could still access it by clicking on the “View as HTML” option which provided the cached web page. When he viewed the Excel document, He Kexin’s name was missing from the list of athletes competing in the Beijing Olympics.
Google’s help page states that “…the cached content is the content Google uses to judge whether this page is a relevant match for your query.”
Next, Stryde searched Baidu, a Chinese language search engine with its own cache and search index. He discovered when he used the search string “site:cn 何可欣 filetype:xls 1994”, which translates to any Excel document with ‘He Kexin’s’ and ‘1994’, it provided two search results with two different excel spreadsheets. Again, when he clicked on the link, the document had been removed.
Again, he was able to view the original cached copy of the document. Stryde explains that the document was available for any user to search and view, without accessing it through Baidu’s cache, until very recently. Stryde poses the question on his blog,
“Do these documents have anything to say about Kexin?” In fact, they do. One document that was located on sport.gov.cn,
clearly lists Kexin’s birthday as 1-1-1994, making her 14-years-old. Stryde claims that these documents are official since they were issued by the General Administration of Sport of China.
Stryde published his theory about the missing Excel documents which alleges that not only is Kexin’s real age 14, but rather, a greater conspiracy is brewing:
“Google’s cached copy of the spreadsheet does not contain Kexin’s age record, and Baidu’s does. This does not necessarily imply that Google allowed its data to be rewritten by Chinese censors, but the possibility does present itself.”
“From the minute I pressed the publish button on this blog, the clock is ticking until Kexin’s true age is wiped out of the Baidu cache forever. It is up to you, the folks reading this blog, to take your own screenshots and notarize them by publishing them. If you put a link in the comments section, I’ll post it.”
Stryde states that his intentions are not to reveal the true age of the gymnast, but to attempt to “demonstrate the power of free citizens to subvert government censorship. The finer points of gymnastics competitions are outside the scope of this post.”
The NY Times recently conducted their own search of these Excel documents. Stryde believes that, “it may have been visits by the NY Times reporter to the official web site that originally caused the Excel spreadsheets to be deleted. I find it unfortunate that the NY Times did not ‘notarize’ and redistribute the primary documents when they were found, if this is the case.”
Hundreds of user comments flooded the blog, most posting links to more screenshots of the Excel documents and some arguing that they were unable to find any results on Google using Stryde’s original search terms.
There are several critics who also posted comments on Stryde’ blog claiming that the name ’He Kexin’ is not unique and that it could be the name of another athlete or that there must have been a typo in the official Excel document. Others state that Stryde should have used other search terms like “1990-1996” or other variations/spellings of her name in order to conduct a more scientific approach.
In order to combat the allegations that He Kexin is a victim of mistaken identity, Stryde ran the original document, still cached on Baidu’s search results, through Google translator and it clearly read:
“799, BB He Kexin CC female AA 1994.1.01 Beijing and
Beijing Beijing Municipal Sports Bureau, First Note”
Stryde ran another search on Google.cn.com searching for the original Excel spreadsheets. His findings were again , odd, because he discovered a new spreadsheet, entitled “05ticao.xls”. After several hours, he searched again and his results were “eye-opening.”
“Expunged entirely from Google.cn’s search index is my original find, zctc.xls. Down the memory hole,” Stryde writes.
For now, Kexin’s real age will be left up for debate. Chinese officials have stated that the 1994 birth date was a discrepancy in the documents. The International Olympics Committee is currently conducting its own investigation. No matter what the result of the investigation is, the rapidly disappearing results from Google.com, Google.cn, and Baidu’s search engine results should be an indicator that Internet users need to stand up to censorship.
It would take hard evidence and confessions to forgery in order to prove that He Kexin is in direct violation of the age limit, and even harder evidence to prove that the U.S. Google’s data has been censored by any government. Stryde believes that digital document forgery is “invisibly malleable” because we lack the tools to track any changes. “The coming wave of remote application providers like Google Docs might someday be able to provide us with a chain-of-trust type solution to this problem, but that day is a long way off.”
Stryder’s evidence has been corroborated by Epoch Times and other news sources, including Kaxin’s own wiki. Kaxin’s wiki documents the removal of several Chinese news articles on the Web that previously reported her age as 13 in November of 2007, and 14 earlier this year. Her date of birth is currently listed as, “January 1, Disputed Year.”
As of Saturday, Aug. 23, the document that was available through the Baidu cache has since been removed.
Stryder claims that he is dedicated to exposing government online censorship and state sponsored fraud.
“For the first time I watched search records shift under my feet like sand, facts draining down a hole in the Internet. Will this stand?” Stryde questioned on his blog.
When Kexin was asked by journalists weeks ago about the debate of her real age, she said, “My real age is 16. I don’t care what other people say.”
Google’s Pay Per Action (PPA) offering has been in beta for several months. As part of the beta, we’ve been running a test campaign on Apogee-Search.com. The traffic volume has been disappointing, to say the least. The campaign has received over 250,000 impressions, and only a pair of conversions. We’re using the same ads that perform well in Search, Contact Match and Site Targeting campaigns on Google.
The last few days have shown a spike in activity; we’ve received five form Submissions that were tagged as Google PPA in our analytics system. The lead quality on these five ran from highly suspicious to out and out fraudulent.
Interestingly, none of the four registered within the Google system, so we weren’t charged for them (a good thing, given their extremely low quality). However, this begs several questions:
1) What determines whether or not Google charges for a PPA form submission?
2) Can the advertiser set any particular conditions of acceptance of a lead?
3) Can the advertiser push back any fraudulent leads?
Numbers two and three are fairly standard characteristics of PPA offerings, and are something Google really should offer.
I asked our account manager at Google these questions, and he responded with the following answers:
1) Although our specific methods for detecting this type of behavior are proprietary, we have some detailed information on the topic posted at http://adwords.google.com/support/bin/answer.py?answer=74490&topic=11637. This article explains our stance on the issue, as well as some of the factors we analyze to detect invalid behavior.
2) Currently, an advertiser cannot set any stipulations as far as what constitutes a valid conversion. From our perspective, if a user reaches the page that contains the PPA conversion tracking code, the action has been completed. As mentioned above, we have systems in place to monitor the validity of these completed actions.
3) The process for reporting potential invalid conversions is the same for reporting invalid clicks. If you notice activity in your analytics system that appears to be invalid, please notify our Click Quality Team by filling out this form:
http://adwords.google.com/support/bin/request.py?clickquality=1&ctx=clickqual. You’ll notice there is a specific checkbox for ‘Invalid Pay Per Action Conversions.’
This, in my opinion, is insufficient. If Google wants to play in the PPA world, it needs to abide by the standards of that world. If an advertiser is going to pay a substantial fee for a transaction, the advertiser should be able to specify the conditions that transaction will meet, and push those fraudulent forms back in an automated manner.
The affiliate world has been doing this for years, has it not?
Stewart Easterby, VP of Sales Operations, Yahoo! Search Marketing
- Gave Panama overview.
- Average of 12-15% of click on Yahoo! Network are identified and not billed for
- Basic API is free. Advanced API (which includes support and marketing programs) is $2,000/month. More info available at booth.
Doug Stotland, Director of Microsoft adCenter, Microsoft
- Want to marry advertisers, but not sure MSN is worthy yet.
- Here from advertisers that the clicks they deliver are good, but there aren’t enough of them.
- Need to make adCenter easier to use.
- Low-quality clicks convert for some people; want to create method for people to advertise on these (and be charged for them).
Frederick Vallaeys, Adwords Product Evangelist, Google, Frederick@google.com
- Showed new Google Analytics and Google Editor interfaces
- New report in Report Center in the next few days: shows what query actually triggered an add, and the match type.
Paul Vallez, Director of Product Management for Search Marketing, Ask.com
- Launching the Ask Content Network. IAC’s sixty brands will be made available.
- Spun the benefits of the overlap between their search and their content networks.
- Can control where ads are distributed.
Question – Is there anything we can do to reduce fraudulent traffic from search partners (smaller search engines)? Many times I can’t even tell which engine it is actually coming from.
Stewart – We monitor traffic from search partners and dump fraudulent traffic. We also discount low quality, but valid clicks.
Paul – We are working with the IAB to define click fraud. The data being sent in from the fraud detection firms is often insufficient or wrong.
Frederick – If you look at the referring URL, you can always tell which traffic is actually coming from Google.
Jeffrey (moderator) – Is there any reason not to be transparent as to where the traffic is coming from.
Frederick – Transparency is good. We’re working on it, but sometimes lawyers get involved.
Question for Yahoo – I’ve not been impressed with the accuracy of geo targeting so far. What do you plan to do about it.
Stewart – That’s not generally feedback we here, but we’re always working to improve. There might be something specific in your case we need to address.
Frederick – Matt referred this morning to problems with AOL traffic all looking like it’s coming from Virginia. We need specific feedback when it’s not working.
Question – I’m a little skeptical about the 90% accuracy claim on geo targeting this morning. What do you think about Matt’s claim?
Doug – I don’t have an exact percentage, but we can have it in the booth. It’s not just the IP address used to triangulate. With the MSN properties, people often sign in and we can get more data.
Stewart – There are three ways we determine a match: 1) a geography in the query, 2) if they’ve entered this into Yahoo as part of a login, and 3) triangulation of the IP address. I don’t have a specific number to share.
Frederick – Accuracy varies greatly on a country by country basis.
Question – Yahoo is generally more restrictive regarding affiliates. What changes do you plan to make?
Stewart – We’re always focused on traffic quality. It’s a balance: advertisers want more traffic, but we need to keep quality high. The affiliate council is an attempt to work this balance.
Question – What is your view of automated bid management tools?
Frederick – Some people have so many keywords that they just need help managing them. Certain tools work well for some customers.
Doug – We don’t think the long term value in paid search management is in actually managing the bids. That said, we don’t think we’re there yet in allowing advertisers to set their preferences. We’d like to have a marketplace where bid management isn’t needed.
Stewart – Advertisers have better things to do than to manage their bids. We’ve encouraged third party tools, and we have our own bid management solution.
Paul – There’s a panel later about whether bid management is dead.
Frederick – The notion is that you give the maximum bid. How many would be comfortable sharing with Google what a click is actually worth (rather than a third party)?
Note – less than ten people raised their hand.
Question – Why is Google’s policy on trademark so loose compared to Yahoo and MSN?
Frederick – Our belief is that if the agency or third party is making a better ad, and it ranks higher, we’ll show that ad first.
Question – For Yahoo, what is the philosophy behind how blended the network is between search and content? How do I identify the traffic properly?
Stewart – It is a blend. It’s anyone we think will drive quality traffic. We constantly measure the traffic and will boot people out of the network. Third party tools may loose some of the referrer data. Our own tools should be much more accurate.
Question – Google and Microsoft now owns two of the top SEM agencies. How are advertisers responding to those acquisitions and what are your plans for them? Is Ask jealous?
Doug – Our acquisition makes sense for access to the ad network and their executive team. It’s too early to go into specifics, but we value these things highly and will be careful to protect that value as they are integrated into Microsoft.
Frederick – There have been lots of questions and the government is taking a look, so I can’t speak to many of the specifics. We didn’t necessarily acquire them for Performics.
Paul – I can speak to the jealousy question. I’ve been saying that the SEMs are innovating faster than the search engines in many cases. We want to work with the SEMs, but can understand the desire to have that in house.
Question – Most of my clients run service based business. It only makes sense to run the ads when they’re in the office. Other than Google, when will you start offering this.
Doug – You can do this in adCenter today. You can turn off ads at certain times, or even leave them on at lower bids.
Stewart – Our new platform allows us to roll out new technology quickly, and we’re hearing this from many advertisers.
Paul – It’s actually not an easy problem to solve. It’s coming, but it won’t be real time out the gate. It may take an hour or two at first, but we’d like to get it to a few minutes.
Question – If you have a content network only campaign, how does landing page quality work?
Frederick – At the moment, quality score does not affect the content network, but it will do so very soon.
Question – I’m in a competitive space, and I can’t bid on Google as their too expensive. Quality score doesn’t seem to be a part of it. $5/click on Google versus $1 on Yahoo and MSN.
Frederick – When you first enter a keyword, we make an estimate of the quality. A new account does not have historical data, so it has to be compared to the whole network.
Question – The Google API has been out for quite some time, but it’s still in beta. What is the defining milestone to take it out of beta.
Frederick – It’s not going away. The structure is where it needs to be. Between you and me, consider it out of beta.
Stewart – To make it clear, the Yahoo program is not in beta.
Doug – I’d also like to say that DOS is officially coming out of beta.
Question – My cost of sales almost doubled with Panama
Stewart – The feedback from Panama is that most people are seeing higher ROI. I’d love to dive into specifics offline. There’re a lot of factors at play. It is an atypical experience.
Questions – Could each search engine outline their policy regarding online pharmacies and prescription drugs.
Stewart – Any advertiser needs to be certified by a third party.
Doug – We take the same approach. Maybe even use the same third party.
Frederick – I believe that we’re the same, but I’d need to put you in touch with an expert.
Paul – We flag pharmaceutical keywords. Advertisers must have a certification on their website homepage.
John Battelle’s Searchblog:
Google, Yahoo, and MSN have announced plans to work together to combat click fraud. Certainly this is a laudable move, but can these fierce competitors work together sufficiently to achieve anything?
And, of course, they each still have a financial incentive to keeping click fraud around.
Associated Press:
Click Forensics reports that the click fraud rate is up 0.4% from three months ago. I’m sure, however, that if we had just let it happen, it wouldn’t have increased.
GoogleBlog:
Google explains that Schmidt’s comments were simply in reference to the theory of click fraud. Even if true, it indicates a lack of understanding of the point of view of their customers.
If anything ever causes Google to truly stumble, you can bet it’ll be connected to their hubris.
ZDNet Digital Micro-Markets:
Google’s CEO suggests that click fraud is a problem that will take care of itself. If a particular keyword (or batch of keywords) is unprofitable due to click fraud, advertisers will stop bidding on that keyword, and the fraud will go away as the bid prices drop.
The hubris of this statement is stunning, but let’s set that aside for now.
Technically, Schmidt is correct when viewing any single group of keywords. However, when that group is no longer profitable for the AdSense publishers, I mean click fraud perpetrators, they will simply move on to the next group of profitable keywords.
So, what you have is a moving wave of click fraud which never ends. Meanwhile, Google continues to make money and the advertisers continue to lose money on the theft.
Spam Blocker Resouce:
So, this guy discovers someone using his wi-fi connection. A packet sniffer shows the guy is using the connection for blog spamming. He shuts the guy’s connection down so the spammer leaves, and he follows the spammer to his house.
Now, he wants suggestions of what do to with this information.
Certainly, nothing illegal should be done, but maybe some vigilante justice is in order. Something in the publicly shaming arena would probably be best.
