Follow us on Facebook

How Traffic Exchange Sites (MGID,Wahoha etc) can Harm Your Google Adsense

Many bloggers have been in panic over the last few months, since Google has been involved in a huge clampdown – banning AdSense on sites that have been part of traffic exchanges. At the outset, traffic exchanges have not been doing anything illegal – they are promising readers to move around the web without getting back to a channel (Google, Twitter, or Facebook). However the interplay of human behavior, along with the model traffic exchanges have created is creating big hurdles for Google – both in ensuring value to its advertisers and ensuring efficiencies of their systems. This article is an attempt to unravel this exciting network phenomenon.

The AdSense policy on Traffic Exchanges

Way back in 2007, Google Adsense blog came up with post recommending people not to get onto traffic exchanges. The post clearly suggests that while there may be a value in being part of them, they “may” cause invalid clicks. Given that many bloggers depend on Google solely for their livelihood – there is mass fear about adopting high quality products that help improve user engagement on their sites.

The Resurgence of Traffic Exchanges

Ajax and the ease of installation of widgets has caused a resurgence of traffic exchanges. Traffic exchanges provide value – both to the publisher and the reader, in getting contextual and new unique users from other sites. Sites like MGID, Wahoha, 2leap have been growing leaps and bounds. Testimonials of people saying their traffic increased by 500% in a week makes one wonder if this is a big ponzi scheme. To understand this, we should understand how traffic exchanges work.

  • You lease out real estate on your articles to the TE (Traffic Exchange) widget. The widget decides to show outgoing links here dynamically.
  • When users click on the TE widgets articles on your page, the TE remembers your site (or a set of pre-selected pages), and promotes them more on other sites.
  • You get back traffic – often much more than what you sent out. Makes you wonder how this can be a non-zero sum game. The trick is that every click on the widget takes a user to an intermediate page – where the user is link baited to click on more than one article. They all open in new windows in the background. One click turned to many. Non zero sum.

Due to the overwhelming importance given to user click behavior, links that get promoted mostly look like this. I know you cannot resist clicking on them. Unfortunately, these are not linked to.

Traffic exchanges promise return traffic to be proportional to the outgoing traffic from a publisher. So a publisher is incentivized to click on many of these links himself – in expectation to get back unique visitors.

Google and Click Fraud

Ever since Google introduced their Pay Per Click program for publishers, publishers have been trying ways to increase their revenues. Long tail bloggers and publishers have been using several ways to click on ads on their sites.

  • Click on an ad on your own article 3 times a day.
  • Invite friends to click on ads on your site 3 times a day.
  • Get into an agreement with other publishers to click on each other’s ads
  • and so on

Google’s value proposition to the marketers gets diluted with bad clicks. Since a long time, Google has invested in algorithms to fight click fraud. Ranging from simple to complex, these algorithms have been effective to a great extent. Some heuristics they may be using are:

  • Look for burstiness of clicks on google ads from the same identity (cookie, IP address, Browser signature) across websites, or on the same site. Such users are tagged as fraudulent.
  • Look for burstiness of clicks on google ads on a single site. This may be concerted effort to increase revenue.
  • Infer other information like the referrer page, and look for lists of pages being created for explicit ad clicks
  • Sense if ads are being clicked by humans or a program (by comparing click to non-click time periods

People (cookies) and sites committing click fraud are routinely warned and removed from the system. In effect, Google made good strides in detecting if clicks on ads are fraudulent and non-conformant with normal human browsing behavior.

Even if Google can detect fraudulent clicks with good accuracy, they are still hurting its bottom line. Every bad impression means Google’s servers were being used up for no return. Detecting click fraud (and not charging the advertiser) hurts Google’s profits. They detect fraud to ensure advertisers get good quality leads.

Bad unnatural traffic to a page is bad for all ad networks. Impressions are served with no follow up actions.

How traffic exchanges undermine the click-fraud detection systems

Traffic exchanges offer traffic in exchange for traffic. They do not have any incentive in stopping fraudulent traffic – since they mostly monetize only on their own intermediate pages. When a publisher installs a TE widget on his site, he senses that the traffic he’ll get back depends on what is outgoing. So he starts clicking on the article suggestions. Notice that every click is going across the web to very different sites.

Fraudulent clicks are dispersed across the web by a traffic exchange

Now if you are a receiving site – there’s a good chance that a good chunk of the traffic you are getting from the traffic exchange is coming from long tail publishers who are visiting your pages in the hope of getting something back. Since they are coming from all over the world, at different times – no click fraud system you have can detect them as fraudulent. They all appear as legitimate to you.

On the receiving end, fraudulent clicks are indistinguishable from normal clicks

Say, you are hosting AdSense on your site. A bulk of the traffic you are getting is from publishers who merely want to contribute a click, and are not interested in reading your content – or click on ads you host. This unnatural behavior ensures that the click through rates of ads shown on your site are very low. Many testimonials on the web have spoken of the bounce rate of traffic from exchanges being extremely high. This traffic often only increases your pageview count and does not contribute to anything else.

If a huge number of pages on the web are part of traffic exchanges, then Google suffers a bad CTR rate on most of its ads. Its ad targeting system penalizes ads that are not clicked upon. There’s a good chance that good ads get demoted because of impressions to bad traffic. Google’s bottom line gets diluted over time.

So, are traffic exchanges evil?

Traffic exchanges stand firm that they provide value to users, by adding diversity of content and ensuring publishers get unique users from new demographics. However, their inability to tackle fraudulent clicking by the long tail publishers makes the ad networks vulnerable to bad quality traffic. The bottom line of all ad networks – whether based on impressions or clicks is lowered by fraudulent traffic. This in turn lowers the monetization for participating publishers. A vicious cycle to death. Perhaps traffic exchanges can change their models to ensure high quality contextual traffic and disincentivize fraud. Perhaps ad networks should not show ads when traffic is referred by a traffic exchange.

No comments:

Post a Comment



Like Us


Donate us For Furture