How to identify fraud in your ad campaigns: top metrics to monitor
Mobile ad spending is growing by 20% this year and is valued at over $75bn, according to Forbes. It’s inevitable that as mobile ad spending goes up, so does mobile fraud. RetailDive reports that fraud rates have almost doubled since 2017 and Juniper Research estimates that ad fraud cost advertisers $19bn in 2018. There are many forms of fraud, ranging from click spam, SDK spoofing and false impressions to faked installs. The effects have been devastating on budgets and data sets.
How to identify fraud in your ad campaigns / Jefferson Santos via Unsplash
Advertisers are finding increasingly limited inventory to spend budget, while demand continues to grow, resulting in a huge opportunity for non-compliant traffic providers to steal from advertising budgets across the supply chain (advertisers, publishers and supply partners). So, how do we tackle this issue?
Set Your Baseline
One of the first places to start in developing a fraud prevention plan is in setting key metrics and benchmarks to monitor. These benchmarks will be unique to each business and market, and there is no one-size-fits-all approach.
For example, in some regions it is normal to receive many installs from one IP, since all phones in the region will use the same IP. In other regions, this would be considered unusual. Another highly debated topic is click-to-install-time (CTIT), and each advertiser will likely have a different benchmark, which can be affected by the size of the app or speed of the internet connection and other factors. As such, it’s important to understand what makes the most sense for your benchmarks in your market and among your audience.
Identify Your Key Metrics
Review performance across a set of key metrics used for determining the quality of your incoming traffic, and set a baseline that makes sense for your business.
Here are some of the key metrics that you may want to consider:
* Industry standards and CTIT distribution: short and long click-to-install time can be an indicator of different fraud types. It can also be considered an issue if it presents itself in a disproportionate amount within a campaign.
* Behavioural data on post-install events: retention rates, for example, can also reveal information on the traffic quality. It’s likely that you already optimize traffic based on these events, in order to achieve the highest performance and block anomalous user behavior.
* Session IP duplicate: watch for multiple conversions coming from the same IP address, it might indicate VPN traffic or fraudulent installs, therefore always monitor for duplicates to ensure you aren’t being charged for potentially fraudulent conversions. This can be an indication in some cases, for example in the US it’s common for users to have the same IP from AT&T.
* Conversion rates per source: the hourly conversion rate is monitored in combination with other compliance metrics such as CTIT to recognize suspicious patterns in the traffic. This ensures that the conversion rate is within your benchmarks.
* Unexpected OS version distribution: check if the distribution of your users' OS and the device looks suspicious in a way that all the conversions are coming from a single device brand. You’ll want to monitor anomalies.
* Blacklisted traffic sources: set these in advance before starting a campaign to ensure your traffic isn’t coming from blacklisted sources.
* Unique identifier: depending on your SDK, you might also add an additional security feature to authenticate the postbacks by adding a unique identifier. This ensures that the conversions are valid to avoid any potential fraud.
Be Proactive and Vigilant
Once you’ve established your metrics and benchmarks, you’ll need to shift your focus to proactively monitoring your traffic for fraudulent patterns and non-compliant behavior in the data. User acquisition managers should work closely with a trustworthy partner who can provide the expertise and tools to identify fraudulent patterns, alert them about any risks, and suggest strategies for optimizing campaigns. As you detect issues or anomalies against your benchmarks, you’ll be better equipped to take swift actions and block traffic that isn’t compliant.
While advertisers like you bear most of the financial costs, everyone in the industry suffers from fraud. It’s necessary for us all to work together and stay educated in order to stay ahead of all the different types of fraud that appear on the market - and adapt our solutions accordingly. Open data sharing and transparency between advertisers and among attribution and tracking firms is necessary to combat fraud. Transparency in the advertising ecosystem helps drive efficiency ac of fraud prevention solutions, leading to an overall healthier industry altogether.
Sven Lubek is the managing director of WeQ