《Conviva:流媒体视频分析应避免的五大误区(2023)(英文版)(26页).pdf》由会员分享,可在线阅读,更多相关《Conviva:流媒体视频分析应避免的五大误区(2023)(英文版)(26页).pdf(26页珍藏版)》请在三个皮匠报告上搜索。
1、5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 1 5 Pivotal Mistakesin Video Streamingto Avoid in 20235 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 2 Mistakes are part of life.Minimizing them is one of the reasons that businesses depend on cutting-edge analytics.That said,there are some issues s
2、o prevalent in the streaming industry that people dont even notice them.This book looks at five of the biggest:In about ten or fifteen minutes of reading,you should have a grasp of these issues.You should also gain some actionable insights about how to minimize their impact on your organization or e
3、liminate the problems completely.Everyone Makes MistakesDepending too much on data warehousesFragmenting experience analyticsUnderestimating 4K UHD migrationsOversimplifying multi-CDN strategiesMaking decisions based on anecdotal evidence5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 3 Mistake 1:U
4、nderestimating the Complexity of 4K MigrationWhether youre doing it for brand hygiene or quality of experience,your migration will face a number of business and technical challenges.Dont underestimate them;deliver on your Ultra High Definition promises.You need to deliver UHD content despite diverse
5、 device capabilities.Business challenges can include everything from quality control to contract compliance.Public adoption of 4K is on track;the trend is led by cheaper devices.Every-user,every-second,multi-dimensional analytics help identify migration pain points and meet various goals including t
6、hose for cost of ownership,QoE and device support.Key Takeaways5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 4 Your current quality control processes should apply well to 4K.The key is to use analytics and isolate patterns.This avoids overgeneralizing in a complex environment,which,in turn,helps
7、you allocate resources efficiently.Gauging 4K Delivery PerformanceComplexity is a given with any type of data migration.Whether the migration involves a new productivity suite or a core business process,the combination of tech and human factors can get messy.With 4K,the complexity comes from a few d
8、ifferent sources.The usual suspects include:Video and audio encoding End-user device capabilities Licensing agreementsYour major technical hurdle is likely to be audiovisual encoding and compression,specifically the level of dimensionality beyond what your organization is probably used to.Youll prob
9、ably have to deliver:H.264 and H.265 4K streams in parallel Multiple sound options,such as stereo and 5.1 Multiple color ranges,such as Dolby Vision and HDRWhy bother?The main reason you have to satisfy all of these different formats is the fragmentation of audience device capabilities.Some users si
10、mply wont have compatibility with certain codecs,color ranges,or audio profiles.Low-level analytics combined with fine-tuned multi-CDN configuration can help you laser-target regional user capacities and lower the cost of ownership of 4K.Theres more:You probably have licensing agreements that bury U
11、HD rights deep in the fine print.Failing to review your contracts before serving content could quickly turn your competitive advantage into a legal risk.Understanding Technical Complexity During 4K Migration5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 5 At this point,4K might seem like more trou
12、ble than its worth.Should you even pursue migration?Trends indicate that UHD is coming to the wider streaming market,complexity or notOne major indicator is that UHD devices started gaining steam a few years ago.Consumers started seeing much lower prices,especially in terms of screens.Cheap UHD scre
13、ens and budget 4K streaming equipment options are obviously good signs in this context.At the very least,this trend indicates that many users have the capability to receive higher-quality streams.Theres a lot that needs to happen for general 4K streaming adoption other than just end-user device avai
14、lability:Target regions need adequate infrastructure and technology to transmit UHD signals.Customers need to choose to buy services that enable high-bandwidth transmission.Service providers,such as ISPs,need to empower broadcasters with tools to maintain QoE.The question then becomes when all of th
15、ese factors will come together and solidify 4Ks place in the market.The answer seems to be“soon.”Evaluating 4K Adoption TrendsTrends indicate that UHD is coming to the wider streaming market,complexity or not.In the case of systemic failures,packet loss can of course seriously diminish QoE in a UHD
16、stream just as it can in Standard Definition or High Definition.That means maintaining an overview of network performance and fine-tuning content delivery networks is still important.Theres more to it,though.Ultra HD 4K is transitioning from a luxury perk to a core service.You need individual data o
17、n viewer-experience impact and exactly what your audience is doing in response to stream degradation to help guide your decisions about service.5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 6 People like sharper pictures more.It seems simple on the face of it.Higher definitions have better opinio
18、n baselines.The opinion scores increase with higher bitrates.It seems like all you have to do is increase your bitrate and serve everything in UHD,but thats not the case.At least,its more complex than that when it comes to tracking the performance of a 4K migration.For example,are your analytics set
19、 up to show H.264 and H.265 sessions separately?These two codecs can(essentially)deliver the same UHD video quality at significantly different network loads:32 Mbps for 4K with H.264 vs.about 19 Mbps with H.265.Youll need to track streaming metrics other than bitrate when delivery cost and a fragmen
20、ted device landscape are in the equation.On a strategic level,customer demand drives the change to 4K.How do you satisfy that demand on a tactical level?Even some of the highest-demand streaming applications,such as live sports,have adopted 4K.In the on-demand arena,the prevailing industry attitude
21、is that 4K is a brand-hygiene feature.Big video-on-demand players have rolled out 4K in various cost-effective ways,such as packaging UHD with premium,multi-stream subscription packages.Is that going to work for your viewers?To answer that,your best resource will be your own data.Looking Beyond Bitr
22、ate Benchmarks:Navigating Multi-Dimensional ComplexitiesRecognizing the Type of Demand for 4K5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 7 When it comes to optimizing your content delivery network,the multi-CDN model is complex and far from failure-proof.Oversimplification leads to a high cost
23、of ownership,inefficient latency mitigation,and late,potentially counterproductive responses to a higher-than-average number of service issues.Multi-CDN cant really be simple,but you can use strategy and analytics to make it easy to manage and configure.Learn how to start balancing the resources of
24、your organization against the realities of these complex network architectures.Mistake 2:Oversimplifying Your Multi-CDN Strategy Multi-CDN is the industry standard,but enterprises dont usually build to take full advantage.Making smart decisions requires lots of data and accountability through closed
25、-loop feedback.Rerouting in particular demands close attention to detail;otherwise,it can cause more problems than it solves.Key Takeaways5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 8 Image Source:https:/ be competitive,your service has to think beyond planning ahead for standard indicators of
26、high network traffic,such as peak usage times and publication or broadcast schedules.You need a real-time view of network performance.Know Everything About the Client ExperienceMulti-CDN architectures are the norm for online media,especially when it comes to applications such as video on demand and
27、live broadcasting.The higher you go in terms of bandwidth requirements,the more essential these types of techniques are to quality of experience.Set Up Real-Time Monitoring and WorkflowsYour ability to analyze and dissect every aspect of the audience experience is directly related to the effectivene
28、ss of your multi-CDN strategy.Here are some general tips on how to manage this analysis:Collect every dimension you can Use AI-assisted services to hone in on key insights Prefer census over sample data Investigate correlations with other key elements,such as end-user devices and video quality5 MIST
29、AKES TO AVOID IN STREAMING VIDEO ANALYTICS 9 The reason you need analytics directly from the client is simple.The only practical way you can currently manage or even understand such a complex architecture is through a full census measurement.Possibly the easiest mistake to make in terms of oversimpl
30、ifying your CDN strategy is depending on unsafe pathways.In other words,a publisher can create single points of failure through which the entire resilient,scalable CDN strategy can fail.Secure Your Routes and Secure Your RoutingLOAD BALANCER MARKET,BY RELIGION(USD MILLION)201620172.62018-e2019202020
31、2120225.02023-pNorth America EuropeAsia PacificMiddle,East and AfricaLatin AmericaImage Source:https:/ dependence leaves all network routes vulnerable to DNS issues,basically negating the failover benefit of the multi-CDN architecture5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 10 One major exam
32、ple is the choice of method for CDN load balancing.Nearly all VoD services could benefit from load balancing,and many streaming applications depend heavily,or even exclusively,on layer-4 load balancing via the Domain Name System.DNS dependence leaves all network routes vulnerable to DNS issues,basic
33、ally negating the failover benefit of the multi-CDN architecture.An alternative in this case would be an API layer behind your web service.Besides removing the risk associated with depending on the accrued DNS layer,this approach would also let you capture more data than a transport-layer-only routi
34、ng method.That in turn would help you make better,more granular decisions.With a closed feedback loop and your granular audience experience data,youll be able to tell whether you made the best possible routing choices.You can then use that information as the foundation for future success,whether thr
35、ough automation,issue prevention,resource allocations,or even vendor negotiations.The multi-CDN approach can increase the resiliency of your network.You have to build it correctly,configure it well,and monitor it in real time.Even then,youll probably still have issues.The key is a closed-loop system
36、 that shows you the results of your decisions.Each routing decision you make has consequences.Track those consequences and use them as benchmarks to compare the results of various alternate decisions.With a closed feedback loop and your granular audience experience data,youll be able to tell whether
37、 you made the best possible routing choices.You can then use that information as the foundation for future success,whether through automation,issue prevention,resource allocations,or even vendor negotiations.Keep a Future-Focused Mindset5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 11 Love them o
38、r hate them or maybe both but theres an industry-wide dependence on data warehouses.Most streaming enterprises depend on them too much when it comes to analytics.Mistake 3:Overdependence on Data Warehouses Its not worth it to relentlessly hack the data warehouse when theres a purpose-built solution.
39、Different underlying logics make a huge difference in efficiency between data warehouses and streaming analytics.Data warehouses remain essential theyre just not the holy grail.Contemporary analytics let organizations dig down into census-level QoE metrics with speed,scalability,and flexibility.Key
40、Takeaways5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 12 Data warehouses are solutions from around 40 years ago.The technology has improved since then,but the fact remains that the basic underlying logic was not built to satisfy all of the core requirements of actionable analytics for a major Vo
41、D enterprise:Real-time(or at least very low latency)observation Complex operations Census-level measurements Continuous reporting Contextual metricsData warehouse technology was undeniably built to support decision-making at the executive level.It continues to be popular because of its effectiveness
42、 in that role.Still,its far from the holy grail of decision support especially in the streaming and publishing arenas.The Data Warehouse:A Decades-Old SolutionData warehouse architecture has its place in nearly every corporation.But where exactly is that place?The answer:Data warehouses belong in th
43、e decision-making toolbox along with a variety of other specialized analytics tools.That way,executives will be able to make use of the most efficient tool for the job.What Do You Expect of Your Data Warehouse?There are certain complex video QoE metrics that data warehouses fail to answer efficientl
44、y.A common example is buffering.Clients experience buffering events for various reasons.Unfortunately,its usually not efficient to take information from a data warehouse and dig down into the types of metrics that could help uncover root causes.Streaming-Specific Questions5 MISTAKES TO AVOID IN STRE
45、AMING VIDEO ANALYTICS 13 How can an operation with millions of concurrent viewers determine how much time their audience spends buffering due to network issues on a single CDN?Executives need to ask complex questions like this to make informed decisions and the exact question might change from one d
46、ay to the next.How can an operation with millions of concurrent viewers determine how much time their audience spends buffering due to network issues on a single CDN?Organizations are operating at a disadvantage when they attempt to tease apart data into usable metrics using a data-warehouse archite
47、cture only.To calculate a single complex streaming QoE metric:There needs to be visibility on potentially hundreds of questions Engineering teams need to build and maintain huge libraries of operators Additional work needs to go into connecting the information to the The logic integral to a traditio
48、nal data warehouse was not made to operate at scale,with low latency,and with that type of abstraction.To perform this type of task,the most efficient possible way would be a system that rolls up the raw data,performs calculations,and then filters down to the relevant segment.Developing Answers5 MIS
49、TAKES TO AVOID IN STREAMING VIDEO ANALYTICS 14 The goal should never be to know everything.Its to get the right insight at the right time the insight that lets an organization outperform the competition.Data warehouses work.They provide insight.So whats the issue?The issue is that,to be truly compet
50、itive,a streaming analytics system needs three things at once:Speed:Any insight available with at least sub-60-second latency Flexibility:Unlimited ability to provide key analytics Scale:Oversight on every session simultaneouslyThats just not practical with a data warehouse.Luckily,theres a no-compr
51、omises solution available.Getting to Actionability5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 15 When push comes to shove,anecdotal evidence is more persuasive than statistics.That applies most to decisions that affect you directly the types of decisions you make every day at the C-suite and VP
52、 levels.You probably use more anecdotal evidence than you realize.Implementing an optimization system can free up your time and get better results.Robust,multi-viewpoint benchmarking is the foundation of actionable roadmaps.Potential for improvement,audience reach,and engagement impact are the three
53、 key factors to look for when optimizing.Key TakeawaysMistake 4:Anecdotal Decision Making 5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 16 Do you make anecdotal decisions?Probably not at least not entirely.Think for a moment about how these factors might impact your choices as a leader:Personal e
54、xperience Word of mouth and rumors Advertising Biased or incomplete samplesYour personal experiences arent universal.People you trust arent necessarily experts.Advertising is only worthwhile when the advertisers objectives align with yours.Incomplete or cherry-picked data lead to a skewed picture of
55、 performance.All leaders know this,but these factors still influence decisions sometimes.The scariest part of anecdotal decision-making is that your choices may work out.The thing to remember is that they wont work out consistently.For consistency,you need a system:Image Source:https:/ of Anecdotal
56、DecisionsOptimizing Your Optimization ProcessValidateRich ContetDigtal Engagement Playbook5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 17 It almost goes without saying that you need goals.However,you also need clarity and specificity in those goals.For example,if you want to increase revenue,are
57、 you aiming for ad revenue or subscriptions?Your key performance indicators will be different for each objective.Continuing the example:You would want more plays and more viewer minutes if you wanted to increase advertising revenue.This is the main step that makes your optimization process predictab
58、le and actionable.You have to connect any business growth or cost reduction to measurable QoE metrics such as video start times,playback failures,bitrates,or buffering times.You do this by linking your relevant QoE metrics to KPIs,which,in turn,you have already linked to your business objectives.The
59、 next step is to determine whats normal for the QoE metrics you selected.Benchmarking leads progress by balancing realistic expectations and ambitious goal setting.Check out the following section on benchmarks for an in-depth discussion.Once you know where you stand,its time to work out where youre
60、going.Plot out the path to improving KPIs based on benchmarks,audience data,and current QoE analytics.Get into the where,what,who,and how of solving any issues.Establish clear business objectivesSelect a set of relevant audience engagement KPIsConnect QoE metrics to KPIs:Benchmark performanceMake a
61、roadmapAnalyze root causes5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 18 This is the execution part of the process.Mobilize your team to resolve the root of your issue,informed by any guidance and insight you developed in the previous steps.Its not over until its over.Confirm that you met the t
62、echnical requirements necessary to achieve your business objectives.Do this through systematic validation of operations.Set new,clear objectives in line with your organizations overall vision of success.Optimize servicesValidate actionsRepeatDrilling down into key metrics is the only way to make thi
63、s process work.Focus on the metrics that matter for your goals and your operations,improve performance,confirm the effect,and repeat.Benchmarking is essential if you want to direct your organizations resources toward practical,impactful goals.The first step to effective benchmarking is to standardiz
64、e your measurements.You need apples-to-apples comparisons.Especially if youre comparing metrics from different sources,you need to execute the same sanitization and standardization processes on all data.Benchmarking as a Core Optimization ConcernThe first step to effective benchmarking is to standar
65、dize your measurements.You need apples-to-apples comparisons.5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 19 Once you have a standardized point of view,you can start looking at things in a more objective way.You will also probably have a leg up on your competition.Personal experience Word of mou
66、th and rumors Advertising Biased or incomplete samplesYour personal experiences arent universal.People you trust arent necessarily experts.Advertising is only worthwhile when the advertisers objectives align with yours.Incomplete or cherry-picked data lead to a skewed picture of performance.All lead
67、ers know this,but these factors still influence decisions sometimes.The scariest part of anecdotal decision-making is that your choices may work out.The thing to remember is that they wont work out consistently.For consistency,you need a system:Image Source:https:/ from Standardization to ActionOpti
68、mizing Your Optimization ProcessPercentage of organizations leveraging analyticsIdentifyingbusiness processimprovementsUnderstanding andimproving customerexperienceInforming marketingand communicationsstrategiesCreating or modifyingproducts and servicesMonitoringthe competitivelandscapeGuiding compa
69、nystrategyDeriving performancemeasurement/ROI metricsData source:DeloitteBecoming an insight-driven organization,20195 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 20 For example,you can see which of your many QoE metrics that impact a KPI are underperforming the industry.From there,you can select
70、 the metrics that:Have the greatest potential for improvement Affect the largest number of audience members Impact engagement the mostThese are the foundations of an efficient roadmap.Of course,the entire optimization process builds on granular data about the user experience.5 MISTAKES TO AVOID IN S
71、TREAMING VIDEO ANALYTICS 21 Traditional analytics look at a fragmented version of events.Viewer experience is a continuous journey that includes non-video aspects of your service.A holistic,start-to-finish view of audience QoE provides better insight when pursuing your business objectives.Key Takeaw
72、aysZoom out from the video viewer.Its time to take a look at how you approach the entire audience experience.Mistake 5:Fragmented Experience Measurement5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 22 A streaming video is a continuous experience.Its a journey.It has a start,there are events that
73、take place,and theres an endpoint.Its easy to equate that video playback with the audience experience.However,when youre trying to make complex business decisions,that equation doesnt balance.The Audience JourneyPublishers used to target the question of audience experience with metrics that came fro
74、m a single source:the video player.You would have plenty of relevant data,but you wouldnt get the entire story.Image Source:https:/ Fragmented Approach2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026120,000100,00080,00060,00040,00020,0000Streaming video subscripti
75、on revenuesUS streaming video subscription and advertising revenues,2010-2026($M)Data compiled November 2022.Streaming subscription video and video advertising totals include subscription and advertising revenues generatedby virtual multichannel services such as Hulu+Live TV,YouTube TV,DIRECTV Strea
76、m and Sling TV.Streaming video advertising revenues are for instream video ad revenues generated by pre-roll,mid-roll,and end-rollad inventory and do not include banner overlays and brand sponsorships.Sources:Kagan estimates;company data;industry data.Kaga,a media research group within the TMT offer
77、ing of S&P Global Market Intelligence.2022 S&P Global.Streaming video advertisingrevenuesThere is so much that goes into a viewers experience.It starts happening even before they tap that play button.Theres a reason executives care about QoE.Knowing it measuring it,specifically is necessary to make
78、good decisions.Thats why its important to understand the big picture of what your viewers are experiencing.What Is Viewer Experience,Really?5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 23 Imagine you run a VoD service and youre optimizing for higher engagement in your ad-supported subscriber tie
79、r.Youve benchmarked your current performance and found that its lacking compared to your competition.Specifically,youre seeing a high number of streaming attempts with some kind of early termination.This is a key streaming metric,but you cant seem to explain the failures with any of your existing an
80、alytics.There are no technical issues or deliberate blocks that would create these events.The app isnt crashing,either,according to your separate analytics stack.You still want to keep eyes on your content to improve ad revenue,and you dont want to get into a trial-and-error situation.Its time to ta
81、ke a step back.This was a common case that led to one of the earlier innovations in analytics:Ad performance tracking.If you had visibility into the pre-roll ad,you might notice that it buffered excessively,had extremely low quality,or even failed to play entirely,blocking playback.All of those even
82、ts could push your audience away from the video and potentially toard your competitorsIf youre already taking census-level measurements of your video experience,youre on the right track.Youre looking at arguably the most important aspect of the user journey.Remember though:Video isnt everything.Its
83、just the most important thing.Each user has their own complex journey.Taking a Holistic ViewLooking at the Entire JourneyVideo isnt everything.Its just the most important thing.Each user has their own complex journey.The fragmented approach to audience experience measurement inevitably leaves you wi
84、th mysteries.You dont want to have to guess when the future of your organization is on the line.5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 24 Image Source:https:/ if youre focusing on video-based KPIs,you can optimize based on non-video behavior.Do fewer clicks through your library result in m
85、ore hours of playback for your ad-supported subscribers?Do viewers of certain sports streams have a higher chance to dig down into your statistics pages?If you had end-to-end visibility on what was happening during each users journey,you could answer these types of questions.Then,you could take acti
86、on that serves those audience sections.This could help you double down on specific target markets,such as live sports.5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 25 Avoid the Mistakes:Upgrade Your Streaming AnalyticsIf youre like most organizations in the streaming arena,you have a system that
87、more or less works.If youve made it this far in reading,you also believe that things could work better.Youre right.Overengineering on top of legacy or non-purpose-built systems increases your cost of ownership.It increases complexity without adding value.Above all,it keeps you from seeing the big pi
88、cture.Theres a better way.Theres a way thats built from the ground up,specifically for your type of operation.You really can get the entire audience experience apps included for every viewer,every minute,every metric you care about.Check out our analytics technology or click here to see Conviva in a
89、ction.5 MISTAKES TO AVOID IN STREAMING VIDEO ANALYTICS 26 About ConvivaConviva is the leader in streaming media intelligence,powered by its real-time platform.More than 250 industry leaders and brands including CBS,CCTV,Cirque Du Soleil,DAZN,HBO,Sky,Sling TV,TED,Univision,and WarnerMedia rely on Con
90、viva to maximize their consumer engagement,deliver the quality experiences viewers expect,and drive revenue growth.With a global footprint of more than 500 million unique viewers watching 150 billion streams per year across 3 billion applications streaming on devices,Conviva offers streaming providers unmatched scale for continuous video measurement,intelligence,and benchmarking across every stream,every screen,every second.Any Questions?Visit or contact Conviva at .