A common question we hear is what our ClickScores mean. In this blog post I want to shed some light on how our scores should be interpreted and present a real-world ad network example.

Whether from an advertiser, publisher or ad network, the Click Forensics scoring engine evaluates every individual click from a paid click stream. Given all the required attributes that describe a paid click, the engine will score each click on a scale from 0 to 1000.

Our ClickScore is a gradient score and, in a nutshell, the higher the score the higher the value of the visitor behind it. Here is a graphical presentation of the overall ClickScore ranges:

To break it down further, a score can be both an indicator of fraudulent or unwanted click activity and a proxy for the likelihood of conversion. Any ClickScore below 100 indicates an invalid click which is typically considered machine-generated traffic or click fraud, while any score above 100 indicates a valid click.

The propensity of conversion increases with a larger ClickScore. A score of 500 denotes a special position in the gradient range as it represents the average odds of conversion based on our training data set of billions of clicks. Scores above 500 indicate a higher than average propensity of conversion. Likewise, scores below 500 denote a lower than average propensity of conversion.

Though our service assigns a score to each individual click, our customers often prefer to look at aggregate scores across a large set of clicks. Depending on customer usage needs, these aggregates can be based on specific sources (e.g. where the traffic is coming from) or on specific attributes (e.g. a particular bid keyword).

For example, our customers like ClickScore distribution charts as a means of traffic quality visualization. In such a chart each individual ClickScore counts towards one of ten buckets. Each bucket represents a score range of 100 points. Below please see a real-world chart from an ad network customer. This chart represents all traffic received from a specific publisher over a the course of a day:

Overall, this publisher can be considered a good traffic source with an average ClickScore of over 500. The majority of the traffic is on the right half of the distribution chart, though there is a significant invalid portion.

For an ad network, such a traffic analysis offers actionable business insights that helps improve traffic acquisition costs from downstream publishers. Further, within a specific traffic source an ad network can identify the invalid traffic referrers to provide input for filtering, routing and pricing decisions. When managing large ad networks with thousands of publishers, detailed downstream traffic information becomes a valuable business asset.

I hope this blog post has helped to shed some light on our ClickScores and also how they can be used by an ad network. If you have any further questions about our products and services please don’t hesitate to contact us.

Ok, maybe I’m dating myself, but “The Gumball Rally” was a classic movie from the 1970’s about an illegal cross-country auto race.  I can’t help but think about that every time I hear the term Gumblar, which refers to a vicious malware attack that seems to be growing and infecting lots of Web sites.

Every time we think that click fraud is going away, or is under control, we discover a new way of perpetrating it that we hadn’t seen before.  Gumblar is propagating malware that is a variant of the ones we talked about several weeks ago when we announced that in the Q1 Click Fraud Index data we saw signs of malicious scripts that we called “zero i-frame attacks.”  In those attacks, the malicious scripts are clicking on links in an off-screen browser, or zero i-frame, unbeknownst to the user.  The links are, of course, paid listings and the clicks generate fees for the site owner.  Click fraud, pure and simple.

Gumblar is a little different in that it seems to actually swap out valid links in search results for new links, thereby tricking the user into visiting a site she never intended to.  That site can be a parked domain or just a list of ads, or the click itself could be a paid click that generates revenue for the site owner.  This is click fraud, too, but maybe even more insidious than the zero i-frame.

Customers often ask why they need to keep using our service beyond the first few months.  Once we’ve measured the invalid traffic, traced it to the source, and mitigated it from future campaigns, why can’t we just declare the problem solved?  The reason is that fraudsters continue to create new ways to perpetrate fraud, and only the eternally vigilant can prevent themselves from becoming victims.

The New York Times ran a feature article this week on click fraud.  Why you ask?  Because, like spam, click fraud is still a big problem for advertisers. The article pointed out that as the economy tilts downward, advertisers cannot afford to waste dollars. This is a good news, bad news scenario for online advertising.

The good news is that online advertising is highly measurable.  Large advertisers that traditionally have been offline are now shifting dollars online.  This fact has contributed to online advertising continuing to grow as traditional media is in decline.

The bad news however, is that this window of opportunity is narrow.  The online advertising community must embrace measurability and enhance trust to gain share of spend from the big guys. 

There was a significant event this week that helped in that effort.  The Interactive Advertising Bureau (IAB) released from draft the Click Measurement Guidelines.  This document, three years in the making, is a great start for our community to come together around standards and enhance trust. Dozens of ad providers are busily working with third party audit firms to become accredited to the new guidelines.  Advertisers will have a way to gauge the level of commitment from ad provides when this list is made public.

Click Forensics was proud to represent advertisers in this process.  In fact, we were the only traffic quality management firm to participate and were quoted in the press release from the IAB.  Many thanks are in order for the 38 members of the working group for a job well done.

Now, we find ourselves at the beginning.  An opportunity exists to build on the foundation laid by the IAB member companies.  Click fraud is going to be a problem for a long time to come.  Progress is being made.  But in order to re-accelerate the growth of online advertising we need more than standards.  We need a community effort to work together to ensure advertisers have confidence that they get what they pay for.  Articles raise awareness, documents create a process and awareness builds urgency.  But ultimately it will take the effort of everyone in the community to get to the day where trust is commonplace and online advertising becomes the marvelous, measurable media it can be.  We look forward to continuing our efforts toward that goal.