In Part 1 of this article, we talked about the current situation of click fraud in pay-per-click advertising and how it has changed over the years, as the search engines have stepped in to help stem the tide of click fraud. Although search engines have complex and effective systems in place to identify and combat click fraud, individual advertisers should still be on the alert and aware of how click fraud may be affecting their business and where it may be lurking in their online statistics. Here are some tips for becoming adept at recognizing the trail of click fraud in your business.
1. Become familiar with your AdWords account’s campaign statistics.
In particular, the Invalid Clicks and Invalid Click Rate columns in AdWords (and similar columns with Bing and Yahoo) will show you the number of invalid clicks and what percentage the “invalid” clicks are of your total number of clicks. Add these columns to your campaign statistics table or download the data for further analysis, viewed for each campaign and by a variety of time periods. Note that if you are looking back in time, this data is only available for periods after January 1, 2006. By looking at these data points over time, you will come to see what is “typical” invalid click percentages for your campaigns and therefore be alerted to any increase in these numbers. Rest assured that any invalid clicks that the search engines identify do not get included in your overall account stats, nor are you charged for them.
2. Learn how an invalid click is usually defined and spend some time examining your server logs to look for questionable traffic/clicks. Although search engines won’t tell you the exact parameters they use to define a fraudulent click, there are some general characteristics. There are also a number of circumstances that may look fraudulent on the surface, but could be entirely defendable and should be worked into the consideration of the situation.
- Repeat clicks on the ad, especially within short periods of time. Although the most obvious, this can be tricky and is not necessarily indicative of fraud. For example, the multiple clicks may be from prospects window shopping online who may return via the original ad for more product information, etc. multiple times.
- Daily budget runs out too early in the day. Although it’s possible that click fraud may be responsible if you notice your daily budget for an ad running out much earlier than anticipated, there could be a number of other reasons why. These could range from items beyond your control (such as breaking news that suddenly makes your product something front and center) to being the expected fallout of an action taken (such as changing the setting of your ad delivery method from an even distribution during the day to showing ads as often as possible).
- Analytics software may record clicks as a user moves through different pages on your website in such a way that they appear as multiple clicks. Be sure to check how any analytics program you use deals with situations such as the multiple touchpoints a visitor may have after arriving on your site from clicking a pay-per-click ad, so that you understand what is behind the numbers presented. If you use Google AdWords, it’s a good idea to use Google Analytics to understand how the traffic and clicks have been identified and interpreted by Google. You can always use another analytics program if you prefer, but you will then at least know how Google sees the patterns in your clicks.
- A web server log may identify referrers in a general manner. For example, the same traffic source doesn’t always mean that the traffic actually arrived from the same visitor, even within a very short timeframe. Traffic may be identified as coming from google.com, but in reality it is a variety of visits to your site from people using Google – but not necessarily clicking on your ad – to get there.
- ISPs may use shared IP addresses for a large chunk of users to keep costs low. For example, the Internet Service Provider may have assigned the same IP address for everyone in a certain geographic area, and if a number of users click on a particular ad from a regional/local business, it could appear that an inordinate number of clicks are coming from the same user.
3. Keep a close eye on conversions. If your conversion rate plummets, realize that fraudulent click activity is only one of the many possible causes. There could be any number of problems with your site’s usability, layout, or functioning that is responsible for the drop.
4. Use of the Google Display Network can result in a decline in ad performance that may look as if clicks are not valid, unless you approach its use intelligently. Remember that if you advertise on the Google Display Network, you need to target differently to ensure that your ads are only associated with websites that are relevant to your product or service. If you haven’t done so, performance of these ads may be poor, making it appear that fraudulent clicks could be the cause, when it is not necessarily so. If you suspect a problem with a particular site, either because of suspicious click activity or just because the ads perform poorly in terms of conversion, use the Site and Category Exclusion tool to remove it from your ad campaigns on the Display Network.
5. Google Analytics can help identify possible click fraud sources in some cases. If you suspect click fraud, one way to investigate and clarify is to use Google Analytics to filter your traffic in regards to geographic region/country and/or language. This may point up unusually high click activity from specific countries that normally you do not see much traffic from and which you do not target. In such a case, you can consider stopping ads going to that geographic area or language.
Although the search engines have invested a lot of time and expense in identifying and dealing with click fraud, there always will be fraudsters that slip under the radar. The problem is that as we’ve become more proficient at identifying fraud, those involved have become more skilled at evading detection and coming up with complex schemes to trick businesses out of money. When fraud reaches the kind of scale as that involved in the clickjacking scheme uncovered in November 2011 referred to at the beginning of this article, it’s gone beyond the type of detection that individuals have access to. The best thing you can do is to take a critical look at your analytics and web server logs for instances or patterns that look out of the ordinary, and investigate any that you find. If you come across evidence of click fraud, submit it to the search engine for consideration. You may have your ad budget refunded if the case is considered valid but, more importantly, you may help stop a fraudster from using their tactics on others.