"Every bid request that we receive has a UA, and DeviceAtlas allows us to read and understand it." Raja Periasamy, Principal Product Manager at Rocket Fuel
Rocket Fuel is a Predictive Marketing Platform that learns from each moment of customer interaction to deliver and optimize media spend to engage, upsell, and retarget consumers across addressable programmatic channels including display, video, mobile, and social—simultaneously. Rocket Fuel’s Moment Scoring technology makes it possible to go beyond the traditional focus on segments. It instead connects with individuals to reach the right person, with the right message at the right time, on the right device.
We spoke with Raja Periasamy, Principal Product Manager at Rocket Fuel. He manages the company’s mobile product, and has worked at Rocket Fuel since 2010. He loves building products based on Artificial Intelligence that enriches consumers' experience with brands.
DeviceAtlas provides UA (user agent) parsing which is crucial to Rocket Fuel’s reporting, targeting and Moment Scoring-based optimization. Device detection allows Rocket Fuel to go beyond segments and demographics, to target individual consumers at the moment of maximum influence.
Rocket Fuel evaluates more than 85 billion bid requests per day from its supply partners from every type of device and OS capable of viewing ad-supported media. In order for Moment Scoring to make sense of this maze of data--to understand what device and OS a consumer is currently using--Rocket Fuel needed accurate UA information on every impression opportunity. Prior to engaging with DeviceAtlas, Rocket Fuel employed a combination of open source and home grown UA solutions that were unreliable and could not keep up with all the new devices that are constantly hitting the market.
Raja took the time to tell us some stories about how DeviceAtlas has changed the way they look at their business, and has allowed what was previously difficult to be realized.
Rocket Fuel clients often request device-level impression and conversion reports, which provide attribution insights on device type, device age, and over 160 other device characteristics. “Say we are delivering 1 million impressions, half to computers, half to smartphones,” says Raja. “Can we tell iPhone from Android phones? Yes. And we often get requests for detailed reports at that device level. The questions usually go like this: Can you tell me media cost per ad, per device? How many consumer actions like ad clicks or site purchases occurred by device? Are people signing up for our advertisers’ services and what devices are they using when signing up?”
Many advertisers approach Rocket Fuel for a variety of different ad campaign needs, sometimes seeking to narrow the targeting of their ads to certain mobile devices. When it comes to reporting on performance of the campaign, Rocket Fuel provides reporting on the devices within that campaign, allowing advertisers to see what was successful, and limit spending on ads on devices that simply aren’t performing. Rocket Fuel also enables targeting by device type (e.g.: Apple iPad, Kindle Fire, Android Tablet, etc…) so campaign precision becomes a new factor in ad targeting. DeviceAtlas powers that narrow targeting.
Rocket Fuel employs a large team of data scientists and machine learning experts. Their job is to develop predictive models that continuously observe consumer online behaviors, with the goal of predicting the value of each impression opportunity, based on specific context such as IP address, browser, device,, and more. UA is one of the most important ingredients since successful application of this data is predicated on data accuracy.
Accuracy is Paramount
Last but not least, DeviceAtlas detection accuracy was higher than with the previous solution, with the number of unknown devices declining to almost zero. Most other solutions provide a “best guess” when an unknown device is encountered, which would reduce the accuracy of Rocket Fuel’s cross device targeting.
“We've been using DeviceAtlas libraries for 3 years. Every bid request that we receive has a UA, and DeviceAtlas allows us to read and understand it. In all scenarios, we want to understand the context of the user in more detail. Anyone can serve an ad, but we need to know how valuable one impression is from another. You sometimes want to combine different properties within predictive analytics, based on how they make sense. e.g.: you can differentiate between desktop web, mobile web and mobile app - you could also combine with high level device type. If you combine them, you can get greater visibility and more powerful information.”
Raja Periasamy, Principal Product Manager at Rocket Fuel