How Google detects invalid clicks Monday, June 22, 2009

If you had landed on this page venturing how google detects invalid clicks, you are at the right place. The mechanism Google uses is as good and efficient as its search and its well mature at this point of time I would say. Because a lot of people tried to make more money by cheating Google, they took all this input to formulate their secret formula to detect invalid clicks. Adsense is one program which earns google a lot of revenue. Here are some things that I think they would take into account to detect invalid clicks. If you are a web master or a user of Google analytics, you'll know this better.


IP Address - If you are connected to Internet you are assigned an individual ip address and if you've noticed carefully and most consumer Internet connections are assigned dynamic ip address and each time you connect to the Internet you have a different ip address based on a scheduling by your isp. If Google can track you by ip address one might think why not disconnect and connect again to get a different ip address. Clever boy! but never forget, its the same isp and you are given ip within a range allotted to your isp. No matter how many times you disconnect and connect to the Internet you will remain in the same range which lets google know that the click is coming from the same range. Now they raise their eye brows and the following factors will let them detect that it is the same user.

Cookies - If you enter any web site with Google ads, they put a cookie into your browser and if it expires only after a year or so. Unless you delete it, its got an unique identifier to identify when was the cookie planted and that's how they identify you uniquely. Repeated clicks from the user having same cookie is just the most dumb way of getting caught. If you delete the cookie from browse each time before browsing, Google knows the click is coming from a new user all the time, when it detects invalid clicks. Got it? There is no way to cheat Google, unless you are a Internet expert or junkie who cheated for a long time, or possibly got caught many times.

Click Patterns - When one does click on own ads, they tend to click only theirs not other ads or not on other sites. which is a really different behaviour from ordinary. When one intends to make invalid clicks they click and try to run away from the site. Moving on to click on another ad. This behaviour could be tracked. Google knows how much of time a genuine visitor spends on a target site linked from ad. Technically this is called bounce rate.

CTR - Google has the statistics of how much CTR does a visitor from search engine would produce.

CTR from Google side - A well performing ad will have good CTR. If one tries to cheat Google, its for sure they might click a non performing ads in a row. which alerts them.

Mac addr - This is the hardware address of your Ethernet card. I'm sure packets flowing a network use this for a lot of purposes including authentication to enter a network in case of corporations. In other case, it may not be the case, stuff like dial up does not have a Ethernet mac address.

Traffic sources - If a site is new or a site penalized by google for illegal content, it probably does not have traffic from search results. When such sites have high traffic, this is something that they might count on.

Location Geo Tracking - With your ip address, Google can target ads to a particular city or even a small town. The same is used to detect invalid clicks. They can easily know where you are located. Its also easy to use proxy site, unfortunately proxy sites are a outdated way of making such invalid clicks, one could easily get caught. The next dumb way of getting caught is using proxy. They have a list of all proxy servers across the world. I need not give a reason for you to believe that.

There are a lot of other factors that Google knows about a particular user like

  1. Network speed and latency
  2. Your screen resolution
  3. Google user's web history
  4. Surfing patterns
  5. A site's user patterns
  6. Points where one user clicks on screen
  7. Points on site where a normal user clicks

Possibly a hybrid artificial intelligence algorithm learning and predicting about users and traffic behaviour on web. We'll never know it. We'll never know if Google uses it for invalid click detection or not. It is impossible to cheat Google's systems. Also, any violation of adsense TOS will also be detected easily.


Disclaimer: These are my personal opinions about the subject.