Detecting and Preventing Click Spamming

Reading Time: 4 minutes

When we are talking about ad fraud, it is mostly perceived as any unreal activity (like bots and farms), and a more dangerous way of committing ad fraud is underestimated- that is stealing your organic users. The effect is not only that you will be paying for your organic users that you will get either case, all your metrics will be chaotic. Once, you could not figure out the organic attribution fraud, any strategy regarding your app will not be on a strong basis. That is, you may think that your organic growth is being decreased, but actually it may be just they are stolen!

Digital ad spending in 2019 will rise worldwide by 17.6% to $333.25 billion according to eMarketer. However, there is a risk that cannot be ignored, which is ad fraud. As digital ad spend rises, fraudsters are stealing much more from advertising budgets. There are many ways for a fraudster to steal organic users.

Click Spamming

One of the very common techniques in ad fraud is click spamming, where a fraudster generates a high number of clicks from real user devices with/without user’s knowledge. Related to the technique the fraudster used, the fraudulent click may occur in two cases:

  • When the user sees your ad, but actually did not click on your ad
  • Or worse did not see your ad at all.

Since Mobile Measurement Partners attribute installs to the channel which provided the last click before an install, click spamming can steal the attribution of organic installs by providing that last click before the organic install.

What does Click Spamming pave the way for?

Click spamming can be conducted in different ways such as;  stacking multiple ads, sending clicks in the background, sending impression information as clicks where ads seem to be clicked even before they are shown, hiding ‘close ad’ button to force users to click, etc.  As one can guess, spammed clicks also do not show any correlation between the time the user is directed to the store and the first opening time of the app.

Effects of Click Spamming

  • Maybe the most important impact of click spamming is the miscalculation of organic installs, where the advertiser pays for an install that would happen anyway. This will cause a misjudgment regarding the organic effort and may result in a loss in the digital marketing spent thinking that marketing efforts for organic are not successful.
  • Since organic users generally show a better in-app engagement than paid installs, click spamming causes advertisers to spend more on these fraudulent sources thinking they are good quality.
  • Not only the advertiser’s money is stolen by fraudsters, but time and valuable marketing data are also stolen too.
  • Last but not least, having networks in the marketing stack without knowing that they are fraudulent is unwillingly harming the other parties with the stolen installs (such as other ad networks or other marketing channels). For an advertiser, spending budget on this type of networks can waste time and budgets, harm relations and cause unnecessary disruptions in the user acquisition process.

Detecting Click Spamming

Detection of click spamming is in the responsibility of both networks and the advertisers. It is up to publishers (and sometimes networks) to choose to engage or not engage in such fraudulent activity. For ad networks side, they should take their own preventive actions to detect and eliminate these publishers from their publisher base, however, networks make a profit from these activities as well and it creates a conflict of interest.

For the advertiser side, the first thing that can be checked is the unusually high amount of clicks, in other words, extremely low click-to-install conversion rate. However, the most effective way of detecting click spamming is to look at the distribution of click-to-install times (CTIT) per publisher. Since click spammers can not control when the app will be organically downloaded after the click, generally CTITs are longer than real non-organic installs and CTIT’s will show an abnormal distribution. However, rejecting every install with long CTIT is not a solution, because real users can be rejected as well.  

At Interceptd, one of the main things that our algorithms analyze is that we expect a positively skewed CTIT among installs where there is a peak in a number of installs with average CTIT and gradually less installs with long CTIT. If there are several peaks, or the peak is on the long CTIT side, this gives the signal for click spamming.

An Abnormal/Unexpected CTIT Distribution

A Normal/Expected CTIT Distribution

How to prevent Click Spamming?

  • As the first step, be sure that it is more than paying unreal installs.
  • Monitoring traffic patterns to fight ad fraud. If you are not getting the conversion you wanted, or you are getting traffic which is too good to be true, it can be an indicator of ad fraud. You can regularly check out the acquired traffic and publishers, to see if there are any irregularities and take action accordingly.

Dollars spent per fraud traffic is increasing and some MMPs still couldn’t find a solution to take down organic attribution fraud.
 Without an adequate fraud prevention solution in place, most of the media spend of advertisers could be wasted or stolen. The best way to stop click fraud is to invest in a solution designed to fight it. Interceptd is here to identify the best channels to spend your marketing budget and increase the profitability of your app.