Last week, we unveiled a new section of our Web site, which explores the enabling technologies behind our complete measurement solutions. I wrote about two of those sections last week, our Video Placement Multiplier and Viral Reach Database. Today, I’ll round out the trio and cover our Video Metrics Engine technology.
The Video Metrics Engine comprises a range of capabilities for measuring how your online audience engages with your Internet video content. Technically, the Video Metrics Engine is a computational grid that receives, processes, and stores information on millions of video streams every day. It’s fed by our publishing partners’ video players, each of which is instrumented with our tracking code that captures every interaction from every viewer in every video. This information is transmitted to the Video Metrics Engine and is aggregated, processed, and loaded into our distributed data warehouse. Once the video player data reaches our data warehouse, our clients can explore it using our Web-based dashboard.
Oh yeah, all of this happens in real-time and at ‘Internet scale’ – across millions and millions of videos. Not surprisingly, the Video Metrics Engine is a state-of-the-art grid, designed to scale out as the universe of videos expands. Accurately describing this can be buzzword-heavy: on the hardware side, we've constructed a shared-nothing infrastructure using multi-core processors, gigabytes of memory, and multiple disks with terabytes of space per node. On the software side, we use parallel processing and distributed data storage to efficiently collect, access, and retain data. The system is scalable, failure resistant, and highly available. Fun stuff, particularly for our technical team.
Our Video Metrics Engine has been with us from the very beginning of the company, and has grown in capability and capacity as we’ve engaged with our customers. In order to produce Internet video audience behavior in its entirety – as it happens – we needed to create the capacity to cost-effectively process all of the data associated with an expansive video sharing network or an exponentially growing viewing audience.
Now that the Video Metrics Engine is deployed, operational, and growing every day, we’re applying it to uncover some of the fundamental truths of how online viewers interact with Internet video content. The fact that lunch time is one of the most popular times for watching online video didn’t come as a huge surprise. The phenomenon of not just initial attention but concluding abandonment (stay tuned!) did. And we know that we’ve only just scratched the surface when it comes to audience behavior measurement.
So, if you’re a video publisher, advertiser, or viral marketer, let me invite you to check out our enabling technologies. And, of course, please contact us if you’d like to learn more!