Streaming is evolving and improving, but video errors are constant
There is one constant that has emerged from the chaos that was 2020, that despite how many new streaming services emerged and how many consumers purchased new devices to view the plethora of new content, every single one of these new (and existing services) experienced countless video errors. Errors in streaming are nothing particularly new, nor will they entirely disappear for the foreseeable future, as the nature of internet streaming is steeped with various issues that need debugging or more often than not, based on problems that may be outside of the control or influence of video services.
Technical video errors come in all shapes, colors, and values
Much like all web-based services and/or pages, and regardless of the back-end Video Player(s) that an organization uses to power their service or platform, video errors are always displayed using an error code that typically comes with a specific error message (ex: 404 – “Page not found”). However, the level of detail that derives from the code and message is very dependent on the documentation that comes with any commercial, native, or open-source player. All video streaming error messages can be sorted into one of three categories:
- Clear – The error and message point to a specific and identifiable error behavior.
- Ambiguous – Often indicated with a “catch-all” message, these error types can be attributed to many different types of error behavior. Although error origin is provided, there is no specific or identifiable behavior(s) subsumed under that error.
- Unclear – As the name suggests, these video errors are completely uninformative and often read as “Unknown”, “Null”, or something similar.
An analysis of the Bitmovin Analytics database indicated that the distribution of error messages and codes were broken down as follows:
It’s important to note that some of the categorizations can be debated based on other contributing factors. Even so, the most striking takeaway is that 20% of errors are identified as “unclear”. Unclear errors are often the most costly to resolve as they cannot be debugged by simply looking at the message and code, and will need significant engineering or solutions manpower to resolve. Even the best Video Player documentation will not help. Given the severity and volume of the unclear category, it will be the primary value when calculating the cost of video errors and the potential monetary savings that would result from properly identifying unclear errors.
The cost of errors for video streaming
Effects on subscription video streaming (SVOD)
In a recent study and article, streaming service Vimeo released the following numbers that define the different reasons for subscriber churn for SVOD OTT services.
These results are based on a multiple-selection survey, and therefore add up to more than 100%. However, as each statement is treated as an “and” response, selections such as “technical reasons” are considered churn rate for the purposes of this paper’s calculations. Given that the other reasons are based on cost and/or content, therefore, the assumption is that errors (or technical reasons) attribute towards 6% of the churn rate.
According to the same report from Vimeo, the average revenue per user (ARPU) across all SVOD service types is ~$15.
*LTV = Lifetime Value
So, with an ARPU of $15, and an assumptive $15 subscription price/service the calculated customer lifetime value (LTV) with a 6% churn rate is $250 or 500 days. The next logical step is to determine how to increase the average customers’ LTV. According to the Bitmovin Analytics industry insights benchmarking data, OTT providers are experiencing a 6.6% error rate across their services.
Based on data points from the “best-in-class” SVOD services using the Bitmovin dashboard, an individual subscriber (or household) attempts 150 plays/month, resulting in 0.33 errors per day. Over an average consumer’s service lifetime, this adds up to an error acceptance threshold of 165 errors over the period of 500 days.
In a perfect world, if a streaming service could remove 10% of their “unclear” or “ambiguous” errors with more accurate information the customer lifetime would increase by 5 days or 1.1%. In monetary terms: By reducing errors by 10% for an SVOD service with 1M subscribers and a $15/month subscription fee, the 5 days increased in lifetime value for the technical churner segment would result in an estimated revenue increase of $160,000.
Effects on ad-based video streaming (AVOD)
Given that there is a full set of additional elements within an advertising-supported video platform to support ad insertions and content protection – there are different error types that come into play, thus a different error calculation model is necessary. The AVOD cost of error model is based on research from S. Shunmuga Krishnan and Ramesh K. Sitaraman who found that viewers that experienced any interruptions to their streaming service were 2.32% less likely to revisit the video platform than a viewer who did not experience interruptions. This additional data point is especially important for AVOD services that depend on maximizing viewership, and thus the number of ads served, through driving regular consumption of their content.
Based on the Bitmovin Analytics platform for AVOD services, 25-50% of users visit the site or service weekly (recurring users), and each unique user generates around 1.5 plays per week, and AVOD platforms serve around 2 ads/play. At the time of this whitepaper’s publication, the price per 1000 ad plays (CPM) for premium content was around $60 and had an upward trend in price in upscale markets such as Germany. In the scenario where a service maintains a 30% recurring viewership and stands to lose ad revenue from 2.32% unique viewers that experience start or in-stream failures, the service would incur a cost of $1.25 per 1000 errors.
Applied at scale and applying the 6.6% error rate from Bitmovin Analytics industry insights for a service experiencing 25M play attempts per week, this error rate would result in an estimated revenue loss of $3,445/week. Unfortunately, removing all errors is not a likely scenario, however, but in this case, reducing the error percentage by 20% already results in a $35,000 revenue increase.
It’s important to note that costs for both SVOD and AVOD errors are highly variable based on the input values (subscription fee, error rate, subscriber count, etc). To find out the potential monetary impact of reducing errors for your specific use case, please visit our calculator at the following link: https://bitmovin.com/demos/cost-of-errors
Find out how to tackle these unknown video errors and to save this information for future use, fill out the form below to receive our complete “Cost of Errors: How to reduce churn with granular data in video analytics” whitepaper.
Did you enjoy this post? Check out some of our other great Video Analytics content below:
- [Blog Post] Improving the Behind the Scenes Viewer Experience for Video Analytics
- [Blog Post] Mitigating the Cost of Errors with Granular Data for Video Analytics
- [Blog Post] Metrics That Matter: Top-Down Error Reporting
- [On Demand Webinar] Tech Talk: Analytics for Workflow Automation
- [On Demand Webinar] Don’t Fly Blind! How to setup video streaming analytics in minutes with any video player