By now, you’re probably used to our blog posts from the Row, talking about an increase in accuracy, coverage for new platforms, operating systems and devices. Often, we can’t share all the details relating to how we achieve performance improvements, but we can share is a peek inside our Device Lab.
As a product manager, the first thing I do when I wake up in the morning is to check my email and catch up on the news. You never know what new tricks Apple or Google have slated for the day. With events like CES, MWC happening next week, it means that the rest of the year is filled with sales figures, speculations, rumors, and confident displays of the “next big thing”. For us, everything boils down to how can we test and support new devices as they come out to help our partners recognize their audiences on ANY device. Based on that, we adjust our algorithms to ensure we can persist our product accuracy, coverage and longevity. It is a constant exercise in research.
5.0 includes major improvements to our performance with enhanced coverage and increased accuracy across different platforms and use-cases
Above is Mike. Mike is our Lead QA engineer and this picture shows him in our Device Lab. Mike is ensuring that our technology can effectively bridge between mobile web and app (recognize a device as the same device regardless of whether the consumer interaction is in mobile web or in app) for some of the latest devices. It is imperative that we perform complex and in-depth regression tests to ensure we are confident that we get the same DeviceInsight IDs for the same physical device no matter who is running our library. Since AdTruth deploys on-premise with our partners, we often don’t find ourselves at the luxury of research data. So one of the challenges is building complex simulations of real life events. Having access to these devices, and with the help of some homebrew tools we are able to provide researchers with a foundation for their munging. With global devices for our major regions, and an almost infinite selection of virtual devices, we are prepared for whatever the next big thing is, whether it be a wearable device for a puppy or the latest gaming console.
Our latest release, which includes major improvements to our performance with enhanced coverage and increased accuracy across different platforms and use-cases (e.g. bridging) helps our partners better recognize and reach their audiences on every device, all the time.
Collision Avoidance will help our partners improve the effectiveness of their campaigns and analytics and overall make smarter decisions
Our research team comes in all flavors – data scientists, statisticians, device engineers, developers. Irrespective of where we come from, we are all driven by maximal device recognition accuracy. Here’s a page from our data scientist’s journal. The goal is to apply the wins from the 5.0 Android accuracy and bridging improvements to some emerging devices and the ability to predict when devices are likely to be of a lower rating. He mounts the research data and after hours of Rplots, and a maze of sqldf manipulations, he has some very good news for us, evident in the big smile he has for us. AdTruth 5.0 introduces a new system which we call Collission Avoidance. By nature of being probabilistic we sometimes have a recognition error. We are developing a suite of features with tools, such as the DeviceInsight ID Rating, to provide our partners with insights into the probability of collision with respect to every DeviceInsight ID, and help them make smarter decision to improve the effectiveness of their campaigns and analytics.
AdTruth is embracing the cloud to make it easier for anyone in the industry to use our technology and make it easier for us to support our partners
Our “always ship early” development team is a force not to be reckoned with. They bring all of our research work together in solid, and effective product. 5.0 gives birth to our latest innovation: DI-Connect. This is a service that lays the infrastructure for AdTruth to become the standard currency for device recognition. Think anti-virus style automated updates for your library, tools to migrate IDs and self-validate a deployment. All our research and effort in improving accuracy, device coverage, and bridging – we want to get it out to you and used in your business as quick as possible with minimal to zero impact to your own development cycle. This is the promise from DI-Connect and you should stay tuned for more exciting features built on top of it is coming out soon.