Ad fraud is a monumental challenge facing the entire digital advertising industry. According to Forrester about $7.4 billion was wasted on display ads in 2016, a figure set to rise to $10.9 billion by 2021. Juniper Research reported that total revenue loss owing to advertising fraud will reach a staggering $44 billion in 2022, rising from $14.2 billion in 2017.
Ad fraudsters are continuously devising inventive new ways to mislead brands and advertisers, hence digital marketers need to stay several steps ahead and integrate robust anti-ad fraud strategies into day-to-day operations. This article explores the different types of ad fraud, and presents practical solutions to help combat it
What is impression fraud?
This malpractice is when digital ads are delivered and counted as viewed, but have not been seen by human eyes – i.e. fake reporting. This is done through a number of ways; fraudsters use botnets to artificially create impressions and traffic. Sadly, human eyes never see these ads because they’re hidden out of sight, often embedded in hidden frames, which is called ad stacking – a technique of hiding an ad below another ad.
How click fraud affects CPC
Bots are used to create massive amounts of fraudulent clicks on ads. For example, if a campaign is hijacked by bots, it then misleads advertisers into thinking that their campaigns are performing well above normal. Fraudsters often create fake mobile sites with fake traffic based on expensive CPC keywords. If a brand makes a purchaseon these fake sites, this then generates fraudulent clicks and advertisers are then led to believe that the click emanated from a real person. According to the 2017 report from The Association of National Advertisers (ANA) bots were responsible for up to $6.5 billion of ad fraud in 2017.
App install fraud
This type of fraud creates fake app downloads or re-installs apps and is conducted through several unscrupulous methods, including:
- Install farms – apps can be downloaded in ‘install farms’. These are rooms filled with mobile phones, where people manually downloaded apps on behalf of developers or publishers.
- Bots – another favourite tool of the modern digital ad fraudster. These are used to create fake traffic, they can increase user base size quickly, but it’ll be with poor quality users who don’t spend much and likely to have poor engagement levels.
App install fraud leads advertisers and brands to burnthrough user acquisition budgets much more quickly, whilst getting deceptive metrics for retention, ARPU or ROI.
How can you detect app install fraud?
Keep your sensors peeled for the following:
- Suspiciously positive results coming from a single source. This can affect critical KPIs over short periods of time.
- Inconsistencies for click and install origins, number of installs vs the hour that they were conducted.
- Bulk installs that originate from sources with the same device ID.
HOW TO FIGHT AD FRAUD
Blacklisting of IPs
If your performance team is reporting substantial increases for app installs and clicks, something insidious may be afoot. Especially if these anomalous positive results are emanating from devices with the same IP address. Create a database to log these bad IPs to help your team avoid them in the future. You can do this alone, or better still, create an open shared database with other companies in your vertical, and even consider working with your competitors.
Work with 1st party data and track post-install events
By using 1st party data, marketers can gain better control and verification for quality of data, plus the ability to spot suspicious patterns earlier on. Working with data from post install events, such as completing an onboarding tutorial can make spotting fraud much easier, since bots find it difficult to simulate human behaviour. If you don’t see any activity post install of an app or game, you can be sure that the source origin which sent the traffic will be fraudulent.
Detect unusual behaviour
Smart algorithms and machine learning can be deployed to fight programmatic ad fraud. Fraud detection isn’t easy for a human to manually accomplish, as it’s tough to manually check a huge dataset of information from hundreds of digital ad campaigns. Machine learning is an automated process that is constantly learning. By defining logical rules, patterns and ways to react to ad fraud, algorithms can be very effective at detecting fraudulent patterns before an impression, click or install is completed.
The best long-term solution for beating ad fraud hinges on the digital advertising industry’s ability to come together to create communities to help fight it. For example, in Spain several high profile companies like Adobe, Rubicon Project, Tappx and IAS have formed a not-for-profit organisation called FAQFraud. Collectively they are on a mission to fight digital ad fraud, offering support and sharing best practices with one another. Other local country ad communities could take note and form similar collectives to tackle programmatic ad fraud on a global level.
Article by Santiago Soengas, Head of Strategic Partnerships at Tappx
Santiago is the Head of Strategic Partnerships at Tappx, the comprehensive one-stop-shop ad tech solution for both user acquisition and app monetisation.
Possessing both an extensive business background and a keen interest in technology, Santiago manages key Tappx accounts and generates multi-million Euro commercial partnerships.
He is also responsible for driving the day-to-day implementation of digital ad fraud strategies across Tappx’s client base.