Every paid click analyzed by the Click Forensics scoring engine is scored based on its "propensity to convert." That is, the likelihood that the click is from a human being with the intention of engaging on the advertiser's site in some way, whether it is to buy a product, fill out a form, or subscribe to a service.

Valid vs. Invalid
The most important determination of click quality is grading the click as "valid" or "invalid." A valid click is one for which an advertiser is happy to pay. It represents a human being in the target audience who clicked on their ad. An "invalid" click is everything else -- a click for which an advertiser or ad network should not pay. Invalid click types include botnets, spiders, double-clicks, out-of-geo, known offenders, spike anomalies, and more. The Click Forensics scoring engine provides more than 15 distinct click types, or "reason codes," for why a click is deemed invalid. Each and every invalid click is assigned a specific click type.

Propensity to convert
The more difficult scoring task is to grade valid clicks based on their odds of conversion.

To provide online advertisers, publishers, and ad networks with insight into the quality of their valid clicks, Click Forensics has designed a scoring system that assigns a value to each click. Simply enough called the "ClickScore," the engine scores each click with a number ranging from 0 to 1,000.

Scores under 100 are deemed invalid, with virtually zero chance of conversion. Valid clicks score in the range of 100 - 1,000, with higher numbers indicating a higher probability of conversion. While all valid clicks have some chance of converting, the ClickScore attempts to quantify these odds and differentiates between clicks that are unlikely to convert (closer to 100) and clicks that have much higher conversion probability (closer to 1,000).

In order that click scores be comparable across different advertisers, publishers, and ad networks, the system normalizes conversion likelihood around the score of 500. Clicks scored below 500 are less likely to convert, and clicks scored higher than 500 are more likely.

The graph shows the curve of ClickScore vs. odds of conversion. Notice that the curve is not a straight line. While a ClickScore of 200 might represent a conversion probability only half as good as a 500, a score of 900 might be four times as likely to convert, or greater. So while low scores don't mean that clicks are worthless, high scores do indicate significantly higher quality traffic.

Average Click Score - A Publisher Management Metric
Because clicks can be grouped and sorted by source -- affiliate partners, syndication partners, publishers, and sub-publishers -- an important new metric emerges called "Average Click Score." The average click score for a publisher is an indicator of the traffic quality from that source. Because the average click score is available in real time (i.e., at all times), it serves as a useful proxy for the "traffic quality" score of major search ad providers, which are received less frequently and often after poor traffic quality has already damaged earnings potential.

Click Forensics ad network customers use the average click score to improve their overall traffic quality by actively managing their syndication partners and publisher networks. For example:

• Early screening of new publisher partners
• Prioritizing syndication partner feeds
• Providing early feedback to publisher channels
• Weeding out poor performing sub-publishers & sub-sub-publishers