Speaker 1: Hello, and welcome to Understanding Audience Experiences on Google Cloud Platform. My name is Saurabh Gupta, and I'm a customer engineer at Google Cloud for media and entertainment. Media streaming has a global audience with wide variances in device types, user locations, connection bandwidths, and user behaviors. Today, I'll be talking about large-scale media analytics for digital streaming and how Google Cloud's products can make it easy to ingest, analyze, and get insights at scale from millions of users watching your content. This demo shows the power of Google Cloud's fully managed enterprise data warehouse, BigQuery, along with Google Cloud's fully managed enterprise data visualization tool called Looker, to explore the media audience experience and answer the question, how many people are finishing their videos? Over the past few days, members of my team have been watching various short movies. Events are emitted from the Shaka player running in the browser being used for these movie-watching sessions. These events are transmitted to an ingest endpoint running on the Cloud Run serverless platform. The events are enqueued onto PubSub, which is a fully managed global event streaming system. Dataflow is a unified stream and batch data processing service based on Apache Beam. Dataflow uses the event stream from PubSub as a data source and inserts the data into BigQuery. Looker is then used to visualize the data and get meaningful insights from it. So over to the demo. Here you can see a simulation of a streaming services video library, just as what a streaming customer might be expected to see in their streaming service. As a user, I can select any movie to watch. Before we begin, let us activate the browser's debug window so we can see the events that are being emitted for analytics. Let me now select one of my favorite movies, Caminandes. The video starts playing in the video page, and a sample pre-roll ad starts before the video. You can see the different ad quartile events being emitted. Let me click on the ad, which takes me to the ads details page, where a customer would likely see the details of the ad they've just clicked on. Let's return to the video and continue playing. As expected, other events continue to fire. Let me mute, and you can see that event being emitted. And let me pause the video. And again, all these events are streaming directly into BigQuery. So let's take a look at BigQuery. Here is where the events would land in a specific data set. I can run a sample query to see these events land in BigQuery in near real time. This query shows me all the events that have accumulated in the past one minute, including the events related to pausing the ad and pausing the video as I issued them on the player. All of these events can be analyzed in Looker using visual dashboards. Let's take a look at that. Here is the media event analytics dashboard that we've created in Looker based on the data in BigQuery. Up here, you can see the total number of events we have logged, the number of sessions we've encountered this month, and most importantly, the viewership funnel, which shows us the progression of viewership from one quartile of the video to the next. So we can see the number of starts, midpoints, and completion of videos. We can also see where customers are watching the videos in a map view. Additionally, we can also see the device breakdown and get session level details. Further down, we can also break down the ad funnel in terms of how many ads were started, how many were completed, and how many were clicked through. We can also get details on a particular asset and drill into that. So here, let's select Sintel and see the details of what this specific asset is collecting from an analytic standpoint. Again, the data that we got at the coarse grain gross level, we can now drill down at a finer level and get details about this specific video in terms of how many events were issued against it, how many sessions, the viewership funnel, the completion rates, devices, ads, sessions, and also geographic distribution. Let us now return to the presentation. As you just saw, BigQuery and Looker provide a unified view of data across ads, players, and other sources of data regarding your streaming platform. BigQuery and Looker offer self-service access to key insights for the various stakeholders, thus allowing better and more timely decision making. And since these are fully managed serverless components, you're freed from the overhead of managing infrastructure and can focus on running your business. BigQuery and Looker make a powerful combination for streaming media analytics, allowing you to unify analytics across ads, content, viewers, and other sources, derive meaningful insights in minimal time, leading to deep understanding of the audience experience for actionable outcomes. Thanks for watching Understanding the User Experience on Google Cloud demo. Please visit cloud.google.com slash solutions slash media dash entertainment for more details and other solutions for media on Google Cloud.