How do you detect click fraud and bot clicks?
We use sophisticated Click Fraud Detection (CFD) to detect and block most fraudulent clicks. They are not counted and not charged to advertisers. The users committing click fraud will not get paid and their accounts will be permanently banned.
We use various means to identify each computer and each publisher, and multiple clicks from same computer will only be counted once, if CFD did not trigger a red flag. If a red flag is triggered, then all clicks on a publisher's account will be audited and if click fraud is found, none of the clicks will count and the account will be banned permanently.
In addition, we use click pattern and clicks/followers ratio to detect fraudulent clicks. Clicks/followers ratio and click time pattern are two important and very effective parameters that are specific to our platform and not available in other types of PPC platforms like Adwords. They help catch most fraudulent clicks and serve as independent verifications of traditional click fraud detection methods.
Any publisher who is bringing in more clicks than the average clicks/followers ratio will trigger a red flag and we will then manually audit the clicks. For example, if a Twitter user with 1000 followers is bringing in 80 clicks for one ad, while the average is 30 clicks per 1000 followers, the account will be red flagged.
Another feature of ads on Twitter is that the clicks fit an average time decay pattern. If any publisher's click pattern deviates from this pattern, it also triggers the red flag.
In addition, we provide 100% transparency to advertisers. Advertisers can download the real time click stats any time and do their own analysis. If an advertiser sees any evidence of click fraud that we missed, we are happy to have the feedback to improve our CFD.
We can't guarantee that we catch all click frauds, not even Google can. But we do have better ways to detect click fraud because of the nature of the Twitter platform.
In addition, we filter clicks by country and filter out known bot clicks, e.g., twitturls.comm, AideRSS 2.0 (postrank.com) , googlebot, appspot.com, and many others are known bots. These bot generates tens of thousands of clicks on our ads. Bot clicks are just crawlers/indexers that are trying to index the links on Twitter, they are not real humans looking at the pages. Obviously they cannot counted. Clicks from these sources are passed onto advertisers but are not counted as valid clicks.
We emphasize to advertisers that conversion ratio of Twitter traffic may be lower than search traffic (unless it is related to Twitter, making money, free offers) because they are not targeting users who are looking for your service. Accordingly, the CPC is a lot lower as well, starting from $0.08/click.
We understand perfectly that for RevTwt to be a viable platform, we have to watch out for the interest of both the advertisers and publishers. Without strict click fraud detection and bot click filtering, advertisers will leave and there is no money to be made by anyone. On the other hand, if we did not honor valid clicks generated by honest publishers, they will leave and there is no one to post ads for the advertisers. We are strict on click fraud to deliver value to advertisers, and at the same time, we make sure all valid clicks are credited to publishers so both publishers and us can earn some money. And we pay promptly publishers who earn their money without committing click fraud.
This FAQ entry has been viewed 10194 times since it's creation on 3/1/2009.