In the last year alone, our fraud detection systems have identified over 48,175 record-level instances of click fraud, and we’re proud to have proactively saved clients over $1,000,000 in associated costs to date. Minimizing click fraud and recouping associated costs are activities best managed programmatically, but we can share some of the basics you can apply to any account.

What is click fraud, and why should you care?

Click fraud refers to all traffic considered “non-human’ or illegitimate. Fraudulent click activity is most commonly carried out by automated bots that click on ads without intent to purchase. The purpose behind the bots varies from monitoring to data mining to malicious intent. Regardless of intent, the following problems can occur as a result:

1. Wasteful spend
2. Hindered performance
3. Erroneous learnings

So how big of a problem is click fraud? According to a study done in collaboration between White Ops and the Association of National Advertisers, Click Fraud cost advertisers over $7 billion in 2016. The study also found that “travel domains had the highest bot rate, with 17 percent of ad impressions being identified as bots.”

There is some good news, though. Click fraud is only a problem when it goes unidentified. Publishers are generally very quick to correct click fraud. So if you can identify it, you can shield yourself from its impact.

How to identify click fraud

Monitoring campaign performance is the best way to catch click fraud. Non-human traffic creates anomalies in data. If you can find the anomalies, you can fix the problem. Here are four of the most common anomalies:

More clicks than impressions

The first one is really obvious, but many people miss it- an ad placement should always have more impressions than it does clicks. In order for someone to click on your ad, they need to first see it and thus generate an impression. This means that even if everyone who saw your ad clicked on it, you would have a 1:1 relationship between impressions and clicks, so when you find instances where clicks are greater than impressions, you have likely identified an instance of click fraud.

Unusual click-through rate

Click-through rate is primarily used to measure the effectiveness of your ad, but it can also be used to spot fraudulent activity. Paying attention to trends in your CTR make it easier to notice abnormal behavior related to click fraud. Setting up a report to track properties with a daily CTR greater than your average rate can help you spot these instances. When you find properties with an unusually high CTR, take a look at your conversion rate. If fraud is occurring, you should see a CVR that is well below average.

Variance between clicks and analytics visits

Another way to validate click activity is to compare data from separate sources. Monitoring variance between publisher clicks and analytics visits, for example, can be a helpful indicator of click fraud. There is a strong correlation between the two metrics. A 10 percent discrepancy can be expected due to fundamental differences in the metrics. However, anything more should be identified. If consistent, this may be a sign of fraudulent click behavior and worth investigating.

Unexpected increase in traffic

Instances of a large number of clicks coming from the same IP address is often a red flag. In addition, be wary of the IP country. While fraudulent activities can certainly occur locally, foreign locales with unusual spikes in click volume are often a red flag, especially if they are resulting in little to no conversions. If you are concerned about click fraud, investigating your analytics data for these types of anomalies is imperative for spotting and understanding a potential problem with click fraud. Having this data available at low levels of granularity will be helpful in this case.

How we help minimize the effect of fraudulent click activity with Koddi

Click fraud cannot be avoided altogether, but there are measures that can effectively reduce the risk to your campaign.

At Koddi, we work very closely and have strong relationships with the top publishers in the industry. In our experience, once instances of click fraud are confirmed, we are able to work with publishers and clients to quickly resolve the situation.

We highly recommend the use of automation to minimize detriment to campaigns. For example, Koddi’s automated bid algorithm allows you to automatically reduce bids on properties that see spikes in click threshold and have a poor conversion rate. Automation is highly efficient in lessening the risk associated with click fraud and can be customized to fit your specific needs.

Click fraud can be a source of frustration in metasearch campaigns, and it is our goal is to reduce that negative impact. When we reduce click fraud, we are able to free up valuable time and resources and maximize return on investment.