In 2015, advertising fraud resulted in over $6.3 billion of damages; the total is expected to have risen to $7.2 billion in 2016 and will likely increase by a further $2 billion in the course of 2017.
GSI can help you protect yourself against advertising fraud in an effective, lasting way – and help to reduce the problem on a global level.
Identify fraudulent impressions through artificial data traffic - created by bots, automatic scripts, Malware, non-human interaction etc.
Recognise fraudulent clicks on the basis of artificial data traffic – created by bots, automatic scripts, Malware etc.
Recognise artificially generated leads and manipulated order assignments – created by toolbars, tracking-pixel, Bots, automatic scrips etc.
We apply algorithms to traffic passing through our system to find out which clicks were generated by fraudulent activity.
Potential rate of detection: 24%
Using the data we have collected and aggregated, we run algorithms to check for discrepancies in the click sequences.
Potential rate of detection: 93%
This option is the strongest of the three: three sequences are aggregated and checked for irregularities by a full range of algorithms.
Potential rate of detection: 98%
Pretty much every Adfraud-tool tries to measure fraud by using Impressions. However, once you have informed yourself, then it is quite clear that this technique only works where no IFrame-nesting occurs. As well as IFrame-nesting, there are unfortunately further concealment methods, which are not traceable for Impressions.
GSI uses a completely different approach.
An example using the three-way tracking:
The tracking starts.
Collection of data
A server-side call is a request from a
computer within a data centre.
These are usually through virtual desktops in combination with proxies, to generate clicks or impressions.
A proxy is a communication interface within a network. He works as an intermediary, whom on one hand receives inquiries, to then establish a connection to the other side using his own address.
The conventional VPN designates a virtual private (closed-loop) communication network. Virtual in the sense, that it is not a separate physical connection but an existing communication network which is used as a transportation medium. The VPN is used to bind participants in the existing communications network to a different network.
Tor is a network for anonymising connection data. It is used for TCP-Connections and can be used, for example, for web browsing, Instant Messaging, IRC, SSH, E-Mail or P2P. Tor protects its users from the analysis data traffic. It is based on the idea of onion routing.
Web scraping (web harvesting or web data extraction) is the extraction of data on websites. Web scraping software can access directly, using Hypertext Transfer Protocol or via a web browser, the World Wide Web. However, while web scraping can be carried out using a software-user, the term usually refers to automated processes, which are implemented by using a bot or web crawler. It is a form of copying, for which specific data is collected and copied from the web, typically into a central local database or spreadsheet, for retrieval or analysis at a later time.
A Crawler (also Spider, Searchbot or Robot) is a computer program, which automatically searches through the World Wide Web and analysis websites. Crawlers are mainly used by search engines to index websites. Other uses of application are to collect Web-Feeds, E-Mail-Addresses or other types of information.
Known attack source are IP segments from which known attacks via DDos, COA or similar have occurred.
Bot fraud is generated by robots or programs called „Bots“, which are often implanted onto computer systems using Trojans. The Trojan visits websites undetected to build up a user-profile. The operators of such Bot-networks can customise click and view-ability rates individually.
Auto script is a new method in the click fraud segment; these are scripts which manipulate the advertisement at delivery which then trigger clicks, for example, in new tabs.
Pixel stuffing is views within IFrames (browser window within a browser window). These are delivered in the background on normal sites additionally to the delivered advertising banner.
Ad stacking describes the superimposing of web banners. Here, only the banner on top is always visible, all others behind it are not visible. This method has been designed to make it very difficult for Ad-Fraud-Analyse-Tools to check the visibility.
Cookie dropping on End-User Site refers to fraudulent Tracking-Pixel, which intercepts visitors of the website and overwrites cookies from other providers.
Is in principle the same as „Pixel stuffing“, in which the landing page is opened via IFrames.
Invisible advertisements are ad-impressions which are located in non-viewable areas. These are often pushed into a non-viewable area using nested Div-TAGs.
By not paying for fraudulently served adverts, you can increase your ROI straight away.
We offer the right options for integration in almost every potential area.
Every five minutes, data collected is entered into our system; after 15 minutes, the first checks are carried out and thorough examinations of the data follow every night.
Our system recognises complex attempts at fraud across all data and learns to identify new techniques automatically.
In just a few easy steps, you can use our API client to feed your data into your system, generate ad codes, or update our system with your data.
All data marked as fraudulent can be used as evidence in courts of law.
Our system offers you the option of feeding data into our system straight from your ad server system. If we don’t have an interface for your system, our developers are on hand to custom-build one for you.
In addition to blocked IPs, we keep a database of over 49 million IPv4 and IPv6 addresses belonging to hosting providers, data centres, VPN services, public proxy servers, and Tor exit nodes. This database is updated on a daily basis.
Advanced filter options and signs of fraudulant activity highlighted to give you a quick and easy overview.
All data is generated as CSV-files and can be used for external filtering with a HTML-reader.
Fraudsters are quick to develop new techniques – and we are just as quick, updating our analysis algorithms on a daily basis and continuously offer new technical possibilities. The development wishes of our customers have our upmost priority.
To contact us, please use the contact form or call us.
Phone: +49 (0) 6452 - 911 36 - 0
DE-35108 Allendorf / Battenfeld (Eder)