Rating of the Recorded Ad Clicks

Calculation of the probability of fraud: How does ADWARY determine whether an ad click is fraudulent?

If you have ADWARY Monitoring set up, Adwary considers each incoming ad click (see Recording Clicks in real time and calculates the probability of it being a fraudulent click. Here different causes and also different methods can be considered, which affect the calculation by a variety of factors.

Basic Indicators

Many differently weighted indicators are involved in the calculation of the probability of click fraud. Adwary constantly updates these indicators and adjusts their weighting continuously (partly automatically).

Some indicators are absolute, i.e. they signal 100% probability for click fraud (entries in blacklists, for example), so that other indicators and their weight no longer matter. However, most indicators only work proportionally and in conjunction with other weighted indicators. If a certain threshold is exceeded via the proportional indicators, Adwary puts the respective user on the blocking filter and transfers the IPs used to the blocking list of Google and Co.

The weighting is constantly adjusted to the new findings. Adwary uses artificial intelligence (AI) and machine learning for this purpose. This ensures that Adwary is always up to date and has a high recognition rate.

Blacklists

  • Worldwide
  • IPs
  • IP ranges
  • Bots
  • Signatures

Bots

  • Worldwide blacklists
  • Signatures
  • Behaviour

IPs, IP Range

  • Worldwide blacklists
  • Internal self-learning blacklists

Internet Service Provider (ISP)

  • Worldwide blacklists
  • Internal self-learning blacklists

Access Signature (+ Device)

  • Certain characteristics from the access data transmitted
  • Used client, device, operating system and other features, among others
  • Internal self-learning signature database

Access Pattern

  • Manner
  • Location
  • Frequency and time behavior in general
  • Even across different devices

User Behavior on the Page

  • Activity on the site
  • Interest in content

Cross-Campaign Detection

  • Characteristics of other own campaigns
  • Characteristics of other own projects

Worldwide Cross-Customer-Detection

  • Characteristics of campaigns of other ADWARY customers worldwide
  • Characteristics of projects of other ADWARY customers worldwide

Users of Adwary benefit from the data on click fraud gained in other advertisers' campaigns and contribute to the specification of this database themselves (cross-customer detection).

Consideration of Characteristics

Many factors are immediately effective at the first occurrence - either absolutely or proportionally weighted - and thus result in an immediate blocking of the ads for malicious users. Some factors need a learning period until clicks lead to blocking. The duration of the learning period depends on the intensity of the underlying click fraud. However, the learning period is mostly only relevant for the individual and less professionally organized click fraud, since little is known about your immediate competitor who clicks your ads. The vast majority is essentially recognized by the cross-customer detection and immediately blocked the first time it is detected in your campaigns.

Some indicators can only be determined with the optional client-side tracking (also Landing-Page-Tracking). For best accuracy you should use client-side tracking. But don't worry, Adwary also works very well only with server-side tracking only. You can read more about tracking here.