Originally published at: Seven ways Bitmovin Analytics can help streaming workflows
Bitmovin Analytics provides valuable insights that can take the performance and efficiency of your video streaming workflow to new levels. By leveraging Analytics, you can optimize encoding profiles, improve player performance, gain deeper understanding of audience behavior, and ultimately, deliver a superior viewing experience while managing costs effectively. This blog post will explore seven ways Bitmovin Analytics can help you to get the most out of your encoding, player, and ad solutions.
- Boost visual quality for viewers
- Lower storage, CDN and overall video costs
- Monitor and debug SSAI workflows in real-time
- Scale live support and observability with custom alerts
- Reduce video start times
- Dive deeper with playback error debugging
- Generate deep audience and user behavior metrics
- Conclusion
- Related Resources
Boost visual quality for viewers
To deliver the highest quality of experience for your viewers, you need pristinely encoded video at resolutions that match the viewers’ screens and bitrates that can be delivered over the public internet. You can make a best estimate at what these combinations should be, starting with something like Apple’s examples in their HLS docs, which is what many people and companies do. What you’ll soon find though, is there is no one size fits all solution, and the ideal settings will depend on your specific content, your audience, their locations and their viewing devices.
This is where Bitmovin Analytics can help shine some light on where you have room to optimize and potentially improve quality for your viewers. The Video Quality section shows useful stats like your audience’s download speeds and video bitrate being played, which gives an indication of their potential for viewing higher bitrate streams.
Another useful metric is the scale factor, which shows the relationship between the playback window size and the resolution of the stream that was delivered. For example if you see the scale factor is above 0, that means your content is being upscaled and could appear pixelated, so you should consider adding a higher resolution rendition to your encoding. The Top Screen Resolutions graph shows what screen sizes are being used to view your content most, so it’s another way to see how much of your audience can handle 720p vs 1080p vs 4K.
Could your viewers potentially handle higher quality than you’re giving them? Is your audience ready for 4K? For streaming services that want to be known for their quality, these are the questions that should be asked and can be answered with Analytics. Bitmovin’s VOD Encoder has the wide codec support and deep configurability that lets you turn these insights into actions.
Lower storage, CDN and overall video costs
Besides the factors mentioned above for the highest quality viewing experience, another part of the reality for streaming services is cost. Encoding, storing and delivering high quality video at scale can get expensive, but reducing costs does not necessarily mean lowering quality. An easy way to find the sweet spot is to use our Per-Title Encoding, which creates the ideal encoding settings for each piece of content to maintain high quality without wasting bandwidth.
However you are choosing your encoding settings, Bitmovin Analytics can add another layer of insight to help you identify where changes or cuts can be made to your encoding stack without impacting your viewers. In some cases you may even potentially improve their experience.
One great resource for understanding your audience’s experience is the video bitrate heatmap. This gives you an idea of which combinations of bitrate and resolution are being seen most often. If you find there are renditions that are rarely being viewed, you might consider removing them from your encoding ladder. Small changes like that can lead to massive savings over time.
Looking at the heatmap below, most of the viewers are seeing the 7.8 Mbps, 1920 x 1080 version, but it appears there are also 5.98 Mbps, 4.88 Mbps, and 3.75 Mbps versions at 1920 x 1080 that only have a few views each. By eliminating one or more of those “extra” 1080 versions from your encoding ladder, you’ll save on the cost of encoding and storing something that is rarely seen. Your viewers won’t know the difference and your budget (and finance team) will thank you.
The video bitrate graph shows the median value across Bitmovin’s customer base. While all content and scenarios have different requirements, this can give you a general idea where you are compared to the rest of the industry and if there’s potential for savings.
Another way you can maintain quality for the viewer while reducing overall costs is by using more modern video codecs. While H.264/AVC is the old, reliable standard that just works everywhere, HEVC, VP9 and AV1 can all deliver the same or better quality at much lower bitrates. That translates to reduced CDN bills, which are usually the most costly component for streaming platforms.
The Video Codec Support graph shows you how many of your viewers could have potentially used a different codec. Adding newer codecs will slightly increase encoding costs, but the longer term savings with your CDN will usually make up for it, significantly so with popular videos. The exact break-even points will depend on your CDN contracts and pricing, so we created this calculator to help you decide when it makes sense to add AV1 to your encoding stack.
Monitor and debug SSAI workflows in real-time
The transition between ads and the main content during server side ad insertion (SSAI) has been a blind spot for troubleshooting and tracking, because the boundary between the main content and ad has been invisible to the player and analytics. This made it especially hard to get to the root cause of errors around that handoff. Now with Bitmovin’s SSAI Analytics, you can get an unprecedented view into your ad breaks, transitions and performance.
Our SSAI tracking collects the relevant data to be able to identify and debug errors like rebuffering, failed ad beacon requests and glitches when switching between encrypted and clear content. Having this level of visibility helps you minimize the cost of errors and playback disruptions.
There are also engagement statistics that show the number of ad plays, plays per quartile, and abandonment rate. Each metric can be filtered to identify commonalities between viewers, devices, stream types, and ad servers. That data can then inform decisions like expanding ad campaigns that viewers are responding positively to or ending ones that viewers are abandoning.
All of this data is available in real time at the session level, including for individual ads within an ad break. This allows for real-time decision making, letting you quickly identify and address issues that might disrupt monetization or the viewing experience.
Scale live support and observability with custom alerts
High profile live events will usually have several people monitoring the stream on multiple devices in different physical locations. That can provide some level of assurance that things are working, but it’s not possible to scale a support staff to actively cover the full distribution of viewers for a major global major event. What about scaling to several or even hundreds of concurrent events? What can you do?
Bitmovin’s Live Encoder is a cloud-native service built to to scale as needed, quickly and reliably. We have customers running thousands of live streams per day through our platform, so we built our Analytics to scale with your event and viewership, multiplying the coverage and effectiveness of your support staff.
You can create custom alerts for a new level of observability across all of your playback sessions. You can define your own alert conditions and rules, including thresholds for metrics like median video startup time, error percentage and rebuffer percentage. You can also filter them by 20+ technical metrics, including custom data fields, so if you’ve been having trouble with a particular brand of smart TV or want to keep a closer eye on quality of service with a specific CDN or in a particular region, it’s easy to create very granular alerts to help you be as proactive as possible.
The alerts can be configured to notify a Slack or Microsoft Teams channel, making sure the right people see it right away, and can get started investigating and resolving the issue before any customer reports arrive.
Reduce video start times
Bitmovin’s cloud partner Akamai has reported that in some cases, impatient viewers will start to abandon videos after as little as 200-400 milliseconds of waiting for playback to start. The big drop off comes after 2 seconds and the losses compound with each additional second of wait time. This is really interesting and potentially scary data when abandonment directly affects your revenue, but using Bitmovin Analytics can help in a few ways.
The Analytics dashboard shows your video startup time across all videos, which you can observe live or across a custom time frame. The startup times can be broken down by video title, geography and several technical metrics for as granular a view as needed. We also provide the median industry startup time as a benchmark to compare against.
DRM load time is another place to check when trying to reduce overall video startup times. It can also be broken down by platform and other categories, allowing you to identify whether slower start times are regional, global or unique to a particular platform or DRM provider. This level of insight allows you to make improvements in minutes and hours instead of days and weeks.
The size of the player itself also contributes to video startup time, so that’s another area to explore if your Analytics data is showing consistently longer than desired startup times. Bitmovin’s modular player can be customized to include only the components you need for quicker loading and better performance. When you need the most performant player possible, our Player Web X offers faster startup times than hls.js and seamless source switching to keep viewers engaged with a steady stream of content.
Dive deeper with playback error debugging
The results of Bitmovin’s 8th annual Video Developer Survey show “Error Rates” and “Buffering/Rebuffering Rates” as the 2 most important video performance metrics to video developers today, by a pretty wide margin.
Conveniently, the Debugging section of the Analytics dashboard lets you zoom in on those metrics and more. Error Percentage shows the number of errors in relation to the number of play attempts on a video platform. It includes errors that disrupted a user’s playback as well as errors from which the player was able to recover. Error Ratio shows the number of errors per hour of playing time. Breaking these metrics down by operating system and player software version is especially useful for identifying a problematic software update that needs to be fixed or rolled back.
Rebuffering percentage is the average time a user had to wait for video segment downloads in relation to the total time a user spent watching a video. Buffering time shows the average buffering duration per play. High rebuffering rates could be investigated by breaking the graph down by CDN or browser to isolate performance issues. You could also experiment with different adaptive bitrate logic in your player to see if an algorithm change improves things.
Another extremely useful view for troubleshooting player issues is the Top Error Codes section of the dashboard. Errors are sorted by frequency, and are clickable, letting you immediately zoom in on the most recent individual viewing sessions that experienced that error.
The session view shows a timeline of events leading up to the error and includes the error message, stacktrace, and Network Explorer that captures the last 10 network requests around the error.
When investigating error reports, a support engineer will often run something like Wireshark or Charles Proxy while trying to reproduce the error themselves, hoping to capture valuable hints from the network and application about what caused the problem. Now, with Bitmovin Analytics, you already have that detailed information captured, from every viewing session where it happened. It is a proverbial goldmine for troubleshooting streaming issues.
Generate deep audience and user behavior metrics
Bitmovin’s AI Scene Analysis generates scene-level metadata including mood, setting, characters, and objects as part of the video encoding process. Streaming platforms can use this contextual info to improve ad relevance, automate ad scheduling, generate highlights, enhance recommendations, and more.
This rich metadata enables a deeper level of content understanding than was previously possible. By combining and correlating that data with audience metrics from Bitmovin Analytics like pausing, seeking, replaying, and abandonment, you can build richer, more meaningful user behavior metrics. These can be used for better, more personalized recommendations that make viewers happier and subscribers stickier. They can also be useful to inform longer term content strategy and velocity decisions.
Conclusion
These seven examples show how Bitmovin Analytics provides critical insights for optimizing your video streaming workflow, from encoding to playback and monetization. Hopefully they provide some inspiration for how you can enhance your viewers’ experience, reduce costs, and gain a deeper understanding of your own audience’s behavior.
Ready to experience Bitmovin Analytics for yourself? Sign up for a Bitmovin Trial today and see how you can transform your video streaming performance.
Related Resources
Video: An Overview on Bitmovin Analytics
Website: Bitmovin AI Scene Analysis