Sift operates at the intersection of mobile and big data. More than 1 trillion mobile ads are auctioned each month on the RTB exchanges from more than 2 billion smartphone users globally. Sift's platform is built for this massive scale.

Sift's platform processes over 200k bid requests per second from our 8 RTB exchange partners across 118 countries. We cross reference and enhance each ad request with Sift's user profile data store, which currently contains more than 900 million profiles and is expanding rapidly. From there, we run through our decisioning algorithms to select the best campaign to deliver to that user. We do all of this in under 120 ms, with a less than 0.1% timeout rate. All impression, click, install, and post-install data is then used to improve our decisioning algorithms in near real-time, so we are always able to make the best decisions for advertisers, while delivering a great user experience for our viewers.

8 RTB Exchanges

200k bid requests per second

100% in-app ads with device ID

3B+ mobile ads delivered to date

395k Android & iOS apps

118 countries globally

 

As mobile advertising budgets for app install ads scale into the billions of dollars, fraud becomes an increasingly complex and sophisticated issue. Business Insider recently reported that fraud could cost app marketers up to $350 million. 

Sift’s platform uses advanced machine-learning algorithms to automatically optimize campaign performance for conversion rate and click-to-install time, as well as offers full support for whitelisting and blacklisting of specific app bundles. Sift’s platform monitors conversion rates in realtime for each campaign and algorithm, automatically optimizing campaigns across inventory sources to improve conversion rate. In addition, Sift’s platform monitors the click-to-install time distribution in realtime for each campaign and algorithm, automatically optimizing campaigns to pause ad spend on inventory where the CTI time distribution is not normal. For example, no installs should come in less than 10 seconds from the click, and the vast majority of installs should come in within 24 hours. Since Sift’s platform runs ads across more than 395k mobile apps, Sift also offers full support for whitelisting and blacklisting specific app bundles to help optimize campaign performance for our advertisers.

Normal click-to-install time distribution

Normal click-to-install time distribution

At Sift, we are on a mission to collect the app data for every smartphone user.

With more than 900 million unique profiles of smartphone users around the world and data on billions of installed apps, we focus on analyzing the connections and links between apps - something that we call the App Graph. We leverage all of this data and advanced AI & machine learning algorithms to sift through billions of mobile ads daily in order to hyper-target
app install ads for our clients.

Sift's App Graph: 900M+ user profiles and billions of installed apps

Sift's App Graph: 900M+ user profiles and billions of installed apps