Case Study: Viber

Viber is a free instant messaging and calling app that allows you to connect with friends and family no matter where you are. Founded by Viber Media in 2010, Viber initially launched its iPhone version in 2010, and released an Android version in 2012.

The app now has over 800 million users who can communicate with one another using several different features including Viber’s VoIP video calls, one-on-one messaging and group chats, sticker packs, and much more. In 2014, Viber was acquired by Japanese electronic commerce and Internet company, Rakuten, for $900 million.

The Challenge

Viber gets ahead of ad fraud

Viber has a strong global footprint, but is still actively growing its user base across Europe, Asia and North America. When apps are growing, they often become targets for fraud and Viber decided to get ahead of the problem. In order to scale its paid activity, Viber decided to enlist an agency that would manage media buying via ad networks. The agency was also responsible for onboarding and connecting Viber to new partners, as well as guaranteeing high quality and legitimate traffic -- ensuring Viber wouldn’t become a target of fraudulent behavior.

The Solution

Viber scaled media spend while maintaining data accuracy

Viber had been working with Adjust for over two years, and when they heard about the Fraud Prevention Suite (FPS) they were one of its earliest adopters. By enabling FPS, all of Viber’s marketing campaigns were automatically protected from the most common and widespread fraud schemes.

Right after activation, Adjust automatically started cross-checking the IP addresses of all incoming traffic against a database of known data centers, VPN providers, Tor exit nodes and previously attributed IP addresses. Any traffic coming from repeated or unusual IPs was rejected and attributed as untrusted and both Viber’s servers and network partners were notified in real-time. This screened out installs that came from untrusted sources, and eliminated simulated/emulated installs that were generated on server farms, or were fed into the internet via VPNs or the TOR network.

Distribution modeling and hyper engagement filters worked to protect Viber from two types of click spamming: low in frequency but high in active user/device count; high in frequency but low in active user/device count.

To fight this fraudulent scheme during attribution, Adjust calculated the statistical correlation between the click engagement and the resulting install. If there was no correlation between clicks and installs for a particular source, Adjust started to deny attribution and both Viber’s servers and network partners were notified in real-time.

“Since FPS protected our interests from the start, we were more confident about allowing them a free reign to onboard new partners and buy traffic. With FPS activated in our dashboard we could be more lenient with the agency knowing the data we received was filtered for mobile fraud schemes. This also meant our retargeting efforts were more efficient since the audience was genuine.” says Moshi Blum, User Acquisition Lead at Viber.

The Result

Viber saved 10% of its marketing budget

“On the first month of onboarding the new network, we received premium quality traffic, whereas once we approved the network and scaled the budget, the network’s traffic fraud rates rose from 2% to a whopping 40%,” says Blum. “This was, surprisingly, something that repeated itself in several of the networks we worked with. FPS allows us to experiment with and test new sources so that we know our budgets are targeted towards legitimate users only.”

The use of FPS also resulted in the following:

  • Most of the rejected installs came from distribution modeling.
  • FPS meant no wasted budgets on fraudulent traffic and meant Viber could acquire users who would convert. On average the company saved 10% of its marketing budget, which could then be reinvested in real users.
  • FPS gave Viber more freedom with the agency so it could experiment safely.
  • Viber has now changed its internal evaluating logic for network quality. They no longer just look at the top KPIs but also include fraud rate as an evaluator. Now, if a partner reaches more than 20% of fraud Viber discontinues working with that partner.

Noya Polliack

Director, CRM & Growth, Rakuten Viber