While You Were Asleep: The Serene Sounds of Fraud

The soothing hum of white noise has become a global household staple for those seeking better sleep and relaxation. White noise apps, which mask noise interruptions by generating sounds over a wide range of frequencies, foster a calming environment to help you sleep. According to a Muck Rack analysis by DoubleVerify (DV), nearly 200 articles over the past year have recommended various white noise apps.

The growing popularity of white noise apps has a downside, though. As the audio streaming industry expands, it attracts those looking to exploit it. In fact, we’ve discovered a disturbing trend — white noise apps are being used to commit audio streaming fraud and divert advertising dollars.

The Growing Concern of Audio Fraud

In the last year alone, DV identified two major global fraud schemes targeting audio streaming: BeatSting and FM Scam. These operations falsify audio traffic using sophisticated methods such as dedicated servers, leading to substantial financial losses. At their peak, these schemes collectively inflicted over a million dollars in monthly damages on advertisers without verification protections in place.

Additionally, according to the DV Global Insights: 2024 Trends Report, in 2023, bot fraud escalated at an alarming rate on streaming platforms, with a 269 percent increase in variants. This contributed to a 58 percent uptick in streaming fraud schemes and variants compared to the previous year.

We saw the same pattern with mobile and connected TV (CTV) — fraudsters consistently gravitate towards emerging channels.

For example, apps like “Deep Sleep” and “Deep Sleep Kids” are developed by the same company and have amassed over 10,000 downloads. Each are rated “E” (for everyone) and hold five-star ratings, according to Google’s download data.

On the surface, these apps seem safe and legitimate. But DV discovered that they generate fake streaming data by selling audio impressions where ads were never actually played. The developer executes this through a method similar to server-side ad insertion (SSAI) fraud found in CTV.

How it Works: Detecting the Fraud

Fraudsters are becoming increasingly sophisticated, using tactics such as spoofing residential IP addresses and mimicking audio apps. They set up counterfeit SSAI servers to fabricate audio ad requests, creating the illusion of active users and inventory desirable to advertisers. These fraudulent requests are then sent to supply-side platforms (SSPs), ad exchanges and ad networks. When an advertiser wins a bid for this inventory, their ad spend goes to waste on a non-existent opportunity. This fraudulent activity not only wastes advertisers’ money but also diverts funds away from legitimate audio channels and publishers.

One way to spot this fraudulent activity is to look at the usage patterns of white noise or sleep apps. These apps typically see more usage during nighttime hours. However, fraudulent apps will show a spike in usage during the day, which is inconsistent with genuine usage patterns. 

The chart below illustrates how a fraudulent sleep app demonstrates the same traffic patterns as other apps, such as those playing yoga music, R&B music, arcade music and exercise sounds. These apps all show a simultaneous spike and fall in their number of impressions, which is a classic signal for fake traffic. 

It’s highly unlikely that the peak usage hour for genuine R&B music apps would be at 11:00 am, and it’s even more unlikely that this would correlate perfectly with the usage patterns of both exercise music apps and sleep apps. The only plausible explanation for this level of correlation is that the audio traffic on these apps are fake, generated impressions that are being fired by scripts or bots.

The chart below shows the behavior of a legitimate sleep in green and that of a fraudulent app in purple.

The DV Fraud Lab detected this audio streaming fraud scheme using machine learning technology and manual review, which revealed that the ads were not being broadcast.

To date, we’ve identified hundreds of long-tail audio apps that are committing this type of SSAI fraud. Annually, the DV Fraud Lab detects over 1,000 apps generating fake impressions and attempting to blend in with legitimate traffic.

Addressing Audio Fraud

Discovering fraudulent activities in the audio streaming sector is a significant challenge, especially when involving seemingly harmless white noise apps. Throughout 2023 and 2024, dozens of apps were discovered in this group. On average, unprotected buyers were purchasing over 45,000 monthly impressions on each app. Even with a conservative estimate of audio CPMs at $5-$7 (many are often over $20), we estimate an impact of at least $225,000 per app, monthly. Given the number of these apps and their popularity, it’s clear why fraud detection is critical.

This situation highlights the critical need for advertisers to be diligent and implement verification and monitoring for audio streaming traffic to protect their investments. For legitimate app developers, fraud like this diverts ad dollars that would otherwise go to them. For users, it underscores the importance of choosing white noise apps from trusted developers — ad fraud on apps can lead to unexpected battery drain and increased data usage. Most importantly, fraud undermines trust in the nascent audio streaming market, threatening greater spending.

As the audio streaming landscape continues to grow and evolve, so too must the measures to combat fraud. At DV, we are at the forefront, working to detect and mitigate these threats, ensuring a safer environment for all stakeholders in the digital ad ecosystem.