The Importance of the MIT Sutdy

There has been a lot of buzz lately around studies regarding click fraud.  Over the past year or so, several groups have conducted studies estimating the size, scope and impact of click fraud.  Click Forensics approach has been different.  We built the largest independent network of advertisers and aggregate their data to determine the size, scope and impact of click fraud.  We believe this is a much better approach and given the growth of the Click Fraud Network, it will only get better.

I was thrilled when Professor Catherine Tucker from MIT called me to propose the study.  Her idea is to focus on sources and motivations for click fraud.  This is an outstanding approach and these type of questions need to be addressed in a thoughtful and thorough way.  The confidential data we will be able to contribute from clients and the members of the Network who agree to participate will provide an interesting look at the problem.  I am looking forward to working with Catherine and her team to uncover more information that we believe will benefit our industry.

Click Forensics has a history of working with the academic community.  One of the first steps we took was to work with Dr. Alex Tuzhilin of NYU to help us determine the statistical significance of our data set.  Alex was also instrumental in helping us refine our scoring methodology to improve accuracy (see blog entry from April 22)). We have inbound interest from other fine universities and look forward to working with the community to provide meaningful data.

In Kevin Newcomb’s article called, “Whose click fraud numbers do you trust?” he was trying to understand our motivation in participating in the study.  We have been consistent from the beginning; the problem of click fraud is an industry problem.  Solving the problem requires a team effort beyond Click Forensics.   That includes advertisers (big and small), agencies, search providers, third party ad serving firms, academia and our competitors.  Over the last few months I have spent time talking to leaders in each of these types of organizations discussing a solution.  We built and fund the Click Fraud Network to provide advertisers valuable information about both their own campaigns and the industry as a whole.  It provides what Catherine describes as “real world data”.   The advertisers, agencies and third party auditors choosing to participate in this study should be applauded.  The questions should be aimed at those behind the curtain.

Tom Cuthbert

The Number One Issue

The discussion around click fraud is no longer an advertising industry topic.  There has been a growing interest from the financial community over the past few months.  I have had several analyst meetings, spoken to numerous reporters and have met with leading advertiser and agency executives.  There is one common thread to these meetings… they are all concerned about the impact of click fraud on our industry.  This raised awareness is a great step forward, benefiting the entire industry by helping to create a positive dialogue between search providers and advertisers to begin partnering towards a solution to the problem of click fraud.

In the past few weeks there have been numerous articles discussing everything from bot attacks to lawsuits.  I have appeared on CNBC twice to discuss the impact of the Click Fraud Index reporting at a 14% threat level.  While this coverage is good for the industry, in my opinion the most critical issue facing our industry is not how we define click fraud, but how we communicate concerns to the search providers.  We believe Click Forensics has the most complete dataset, the most advanced algorithm and the most accurate approach to identifying unwanted clicks.  But what can an advertiser do with that information?

Currently, just about all they can do is send in log files, excel sheets or hope the ROI on their campaigns will improve over time.  It is this situation that is causing the growing frustration in the industry between advertisers and the search providers.  In other media (TV, radio and print) standards exist for advertisers and publishers to agree on delivery.  The answer for the online community is to establish and agree on a way for advertisers to submit their concerns to the providers.  Becky Quick from CNBC asked me if I thought there would be a deal in place before the end of the year.  The answer is yes.  Stay tuned as I will have more to say on this topic at the Bear Stearns Internet Roundtable in New York next week.

Tom Cuthbert

AdTech Update

The past couple of months have been really exciting for all of us at Click Forensics.  Exhibiting at the AdTech show in San Francisco was a great success and over 500 new members have joined the Click Fraud Network.  We now have over 1,200 advertisers, agencies and search providers as part of the Network.  I thought you might be interested in some stats about our members.

In the advertiser group, 8% of the members are Fortune 1000 companies and several are among the largest advertisers online.  25% of the members are in the retail space, 16% in financial services and another 11% fall into the technology category.  These vertical markets, along with others, will give us a basis to study in the coming months.  There is a clear difference in each category as to the amount of click fraud that is occurring.  Look for the Click Fraud Index to release information on this in the coming months.

AdTech also was the launching point for our Click Forensics 4.0 for Enterprise™ product.  This hosted solution is geared for advertisers with over 100,000 paid clicks per month.  We have had terrific interest in this area and are proud of the product.  It is a rich reporting interface and a very complete solution for the problem of identifying unwanted clicks.

Tom Cuthbert

Mark Cuban: Agree or Disagree

I first met Mark Cuban in 2000 right after he bought the Mavs.  He was speaking at an industry conference in Dallas and was talking about “stupid technology company names”.  We had just launched a company and when I told him our name, he laughed and approved.  Five years later, when we named Click Forensics, I recalled his advice and tried to avoid naming it after a Greek god or some other random seven-letter name that happened to be available.  He has yet to comment on our name specifically,  but he did mention Click Forensics in a recent blog on click fraud.  While he never fixed the dead link to the Click Fraud Index, he did have some interesting takes on the topic.

Cuban said, “The concept of “I know some percentage of my PPC advertising is click fraud, I just don’t know home much” shouldn’t be acceptable.”  We agree and that is why we have created a company solely focused on solving the industry problem of click fraud.  In his more recent post, “Why I think click fraud is far greater than imagined”, he boldly states, “…no amount of IP repetition algorithms are going to stop them.”  Once again we agree.  In fact, nothing is going to stop them.  The solution is police the activity and report on discrepancies.  Once these unwanted clicks are identified, there needs to be a way to reconcile them so advertisers aren’t stuck with the bill.  Keep blogging Mark, you have friends in high places.  This is an industry issue that needs guys like you to speak the truth.  

So what do I disagree with Cuban on?  Well, Click Forensics is based in San Antonio; we have season tickets, know several players and even have a room in our office called the “Spurs room”.  I grew up in Dallas but left before the Mavs were born.  While I root for the home team, Cuban is good for Dallas, good for the NBA, good for the online community and even good for San Antonio.  Next year…

So Mark, after you get your ring (and you will) drop by to see us in SA.  We can celebrate your victory on the Riverwalk, eat at a Dairy Queen and talk click fraud in our Spurs room!  Ciao~

Dr. Tuzhilin Visits Click Forensics

Click Forensics has been working with Dr. Alexander Tuzhilin from NYU to further enhance our rating engine and statistical modeling approach. Dr. Tuzhilin visited with the Click Forensics’ team in San Antonio last week. The company has a goal of producing reports with the highest degree of accuracy in the industry.

[caption id="attachment_693" align="alignright" width="248" caption="Dr. Tuzhilin works with the Click Forensics executive team at the San Antonio headquarters"][/caption]



Dr. Tuzhilin said,

"Click Forensics has good data and this is a source of their advantage over the search engines. My role is to work with them to refine the scoring methodology to improve accuracy. Their approach is to incorporate as much data as possible to improve accuracy. The search providers simply don't have enough data to have the most accurate approach."

The work of Dr. Tuzhilin has helped Click Forensics build a road map for the future.  We appreciate his insight and expertise and look forward to working with him for years to come.

Click Fraud Network Update

Since the launch of Click Fraud Network on March 1^st, the Network has grown rapidly.  Literally hundreds of of the top advertisers in the pay per click space have joined and we are adding members daily.  This is important for two reasons:  First, click fraud continues to be a serious problem.  A recent CNET article noted that despite the proposed settlement with Google, click fraud is not going to go away anytime soon.  Secondly, there is strength in numbers.  As our Network builds, so will the clout of our members in working towards solutions.  As I have pointed out for some time, the problem of click fraud is bigger than one company or search provider… it is an industry wide issue.

I want to take a moment and provide you some feedback from the Click Fraud Index and share some recent updates.  The overall “high threat level” click rate is leveling off.  As the number of advertisers increases in the Network, we are seeing this number settle down to around 14% of all clicks from all categories.  Tier 1 search providers have a significantly lower threat level than those in Tiers 2 and 3.  This would make sense because they have more internal resources to focus on the problem.  While this is good news it still means that click fraud could account for over $750,000,000 in 2006.  The search providers simply do not have the data available to resolve this issue.  There is a growing call for an independent third party to validate clicks in the same way the Nielsen, Arbitron and the ABC exist in traditional media.  We will be commenting more on this topic in the coming weeks.

Another interesting development has been the spyware issues relating to affiliates sites as pointed out by Ben Edelman.  (His comprehensive studies have validated what we have been watching for a long time)  Affiliate networks can be a risky place to advertise without a monitoring tool in place.  This issue will grow more complex as time goes by and technologies advance.  The exposure and opportunity for click fraud using spyware, bots and organized “clickers” is a real threat that is increasing daily.

We are thrilled by the response to the Click Fraud Network and that so many of you are taking advantage of our CF Analytics tool, offered to advertisers at no charge.  I hope you will encourage others to “Join the Network”

Plan to visit us at the upcoming AdTech show in San Francisco April 26^th to 28^th.  We are exhibiting at booth #6366.  I am also looking forward to being part of the Click Fraud panel discussion lead by Jessie Stricchiola on Thursday April 27^th at 3:15 PM.  I hope to see you all there for some lively discussion!

Best regards,

Tom

The Real Problem with Click Fraud

I have been reading Henry Blodgett’s blog Internet Outsider for the last few months.  His most recent post is called, “The Real Problem With Click Fraud”.  It is well written and right on, for the most part.  It raises an interesting point (that he has mentioned many times) that there is no real data in defining the click fraud problem.

The core of the issue is indeed that “no real data” exists.  How big a problem is click fraud and is it growing? While we don’t claim to have all the answers I do believe we have a more statistically significant dataset than anyone else published at the Click Fraud Network site.  Our member base has grown significantly since the Network was launched a few weeks ago.  We are reporting that the instance of “high threat level” clicks is currently at 16.4%.  As our Network grows we will be able to announce threat levels by vertical market, search provider and other relevant subsets.  The industry will see that there is a big disparity across industries and providers.

Looking at individual campaigns or advertisers and trying to draw conclusions is pointless.  Our approach is to aggregate the data by providing our tool at no cost to the 90% of PPC advertisers who have less than 100,000 paid clicks per month.  This “collective intelligence” provides protection for them while helping the industry by revealing the scope and nature of click fraud.  Similar to the approach taken by spam filters or anti virus tools, the Click Fraud Network data powers the rating engine making our algorithm as accurate as possible.

Henry is on to something.  We want to help answer the questions.  Click fraud is not “immaterial”, so rather than speculate, let’s work together to peel back the onion.  Advertisers should Join the Network and let’s solve the problem.