March 23, 2012

Preventing Click Fraud

By Delynn Ho, Vice President, Sales

If you're an advertiser, you've probably heard about 'click fraud,' which occurs when a person or automated programme imitates a consumer by clicking on banner ads with the intent of generating bogus ad charges, which in turn can affect your advertising campaign and cost you money.

Late last year, for example, the US FBI busted a cyber-ring that had infected 4 million computers in 100 countries and netted more than US$14 million by manipulating online ads.

But as a mobile advertiser, to what extent is this an issue you need to be concerned about?

And what are companies like BuzzCity doing to protect you?

What is the extent of click fraud?

First, let me put this right out in the open:

1. Fraudulent clicks account for less than two percent of all clicks on the BuzzCity Advertising Network.

2. Our filters detect and discard most invalid clicks before they reach an advertiser's account.

3. While the percentages are small, as the size of the mobile advertising market continues to grow, the dollar value of this type of fraud should not be ignored.

Your best course of action is to actively monitor your advertising campaigns. By tracking your results and focusing on ROI, lost revenue from click fraud will be minimised.


Preventing Click Fraud: What you need to do

1. Track your ad conversions and take note of any unusual activity.

We can place a cookie on a consumer's mobile phone when she clicks on your ad, then retreive the cookie when she arrives on a mobile conversion page that you've created.

The conversion page should be linked to your campaign's objectives. For example, if you're running ads to directly generate sales, the conversion page could be a purchase confirmation or receipt. If you are driving brand awareness, this page could simply be a thank you note with more information about your campaign.

The conversion tracker does not directly detect fraud, but it does help you identify the best performing banners and publishers. If you notice an ad that has a high click rate but low conversion, drop that ad in favour of another better-performing one.

As Kok Fung wrote last month, a good mobile media planner should be doing this anyway in order to maximise your campaign's effectiveness. Plus the clickstream data helps us improve ad scheduling so that your ads appear on the best performing publishers.

Here's an example of a well-performing, as viewed in a BuzzCity Advertiser Spending Report:



And here's an example of a questionable or poor performing ad:



In both cases, first take a look at the conversion column.  Just eyeballing the charts, you'll notice high conversion numbers in the good example and low ones for the poor performing ad.  Next, take a look at the clicks to get an idea of the ratio of conversions to clicks.  Pull out a calculator or excel sheet and you'll find that over a six day period from 13 - 18 March, the good performing ad had a conversion-to-click rate of 81%, whereas in the second example the ratio is just 2 percent.


By the way, you can also sort your data by date, country and campaign. To find out more about how to enable conversion tracking and monitor the results, check out this guide.

2. Track offline indicators

In addition to monitoring online conversions, brands can also track offline indicators, such as sales. If your sales grow during the period of an ad campaign -- and sales growth is your campaign objective -- that's a good sign that the campaign is working. However if sales don't go up, it's time to re-examine your creatives and ad placements.

3. Report suspicious activity

Maybe you've seen a site where your banner experienced a huge jump in activity but almost no follow-through by consumers or perhaps something else caught your eye. If you suspect that clicks from a certain site look suspicious, let us know and we'll check it out.


Protecting your ad dollars: What BuzzCity is doing

BuzzCity's algorithms -- which are based on browser and carrier information, IP addresses and user sessions, among other variables -- allow us to monitor potential fraud. These filtered clicks may show up in advertiser logs but not in your bills. In those cases where fraud is discovered later, we reverse the charges.

Still, there are a few highly sophisticated frauds that require more advanced solutions, so BuzzCity is partnering with some of the best professors at the National University of Singapore to develop new solutions. Experts in statistics and Artificial Intelligence are studying click streams from our campaigns to develop better fraud detection algorithms.

It may never be possible to know for sure the intent behind a click, but research like this will help create mathematical models to better identify quality traffic for our advertisers and improve earnings for publishers.


Related Articles

Learn more about BuzzCity's Publisher Fraud Policy