There was a time during the early 80s, when people were accustomed to CRT TVs covered with shutters to protect the screens, and had very limited access to video entertainment. This changed dramatically over the years, and we are now living in a digital-first era, where everything is connected to the internet.
These past decades transformed the way media is consumed. A lot of media consumption is happening over the connected devices, ranging from small-screen devices like mobiles to gigantic connected TVs.
When it comes to consumption analytics, media companies and brands used TAM reporting to understand the most popular shows, viewers’ consumption patterns, and an estimated viewership based on limited sampled data. Business and strategic decisions on advertisement spends were mostly based on this data collected by the rating agency.
Over the years, BARC took over as a more standardized way to measure the television ratings.
These agencies, however, providing the business insights relating to consumers’ behavior, were meant for traditional broadcasting mechanisms only and offered limited data, which was extrapolated for the analytics purpose to understand the viewing pattern better.
The evolution of digital analytics
The era of personal computers and their use in day-to-day business, along with internet surfing, brought new opportunities in the world of analytics, where the analytics data could now be collected more accurately from each individual.
Google Analytics became the first popular analytics solution that was targeted for the web and offered many insights to businesses. Not to mention, it became the most used analytics platform globally, and found ways to be used as an analytics tool for OTT services as well.
As the platforms extended from the web to mobiles and eventually to CTVs, many analytics platforms emerged offering end-to-end analytics to businesses; some of them even offered cross-platform analytics for data consolidation. Later, some new platforms offered specialized analytics, viz., network performance, player analytics, etc.
The introduction of ML and AI brought new possibilities with them, where data models were built to understand data by understanding the relationship between patterns and variables, and perform predictive analysis.
Challenges with traditional approach
In practice, OTT businesses specifically, put enormous effort working on content acquisition, technical means of delivering it to the end-user, and relative costs; but, they often miss the importance of having an analytics framework that suits their business and a powerful BI (business analytics) solution, which can realize the framework for them.
These are the ones that take the high road of just choosing GA/FB by default without any thoughts in mind, and use the predefined reports for them, which may or may not give them the insights they may look for. That being said, it is not that GA/FB are not the good ones, they indeed are one of the best analytics tools if used to their full potential, but it is about having the understanding of what data points businesses need to collect and how those should be analyzed.
Typical challenges with this approach:
- Lack of analytics framework. While different platforms have their own way of collecting predefined data points, and then uniquely representing them using their dashboards, the major challenge businesses face is not to configure their SDKs and use their dashboard, but, more on what data to collect and what it may potentially offer to their business.
- More often than not, businesses see what the tools may potentially offer them by seeing the analyzed representation of specific data sets, but they miss the point that each business and domain being unique, one solution may not fit all.
- Data retention. Another major challenge is data retention. All major platforms limit the data validity to a specified period for the free tiers including GA and do periodic deletion of data based on specified periods.
- While it looks okay from a cost perspective, it costs businesses a lot by failing to analyze long-term data to do predictive analysis, and see the long-term variation in consumer behavior, service usage, and consumption patterns.
- A typical example is cohort analysis, which depicts a user’s affiliation with the service. With data limit being set to a couple of months, any analysis, which requires the understanding of the user’s affiliation for a longer-term, would become impossible.
- Multiple tools. A lot of businesses use multiple analytics tools to understand their service performance, and this takes too much effort to consolidate data and derive meaningful insight.
- For example, many businesses prefer using OVPs (online video platform) analytics dashboard to understand their content performance, UMS (user management system), and billing engine’s analytics dashboard to understand the subscription performance and GA to see behavioral flows and more user-related data. This makes it difficult to analyze the data analytics journey for a user in a connected flow.
- Cost. It is important to understand how much it would cost to have the right analytics solution in place, and this decision should not be driven purely from the cost perspective.
*Businesses should weigh the decision by comparing the needs outlined in the framework with the tool that can help visualize the insights for them, their data retention strategy, and from the consolidation perspective.
The essence of BI
While it is important to look at the content, technology, and cost for a video streaming business, it is equally important to understand the need for implementing a powerful BI solution that provides the required insights, which can provide logical reasoning to any possible business decision.
For example, in a country like Myanmar, login trend is that more than 80 percent of the users use Facebook login to access the content, which outlines the need for implementing this as one of the preferred ways before other options.
Another example could be understanding the content popularity. For businesses operating across countries, providing multilingual content from different partners, it is important to understand the content popularity by geography where such outcomes can be used to drive further consumption. A recent example of such a case is Netflix. Insights on the content popularity versus content provider would help with the decision making on the content-acquisition front.
It can be argued that these can be provided by OVPs or at the API log level; however, mapping this content with user journey and to uniquely identifying user clusters for targeted campaigning or personalization requires using a single analytics solution to capture various data points.
The solution is to have an analytics framework that maps the business needs with technical detailing. It acts as a master reference, where the business KPIs are broken down into smaller insights that a business may need, and maps them to properties/metrics/dimensions/parameters that should be tracked at the front-end applications.
It should also contain how the reporting should be done, and how these tracked parameters are mapped into different reports such that they provide required insights to the businesses.
The next step is to devise a data-retention strategy, which involves getting answers to questions like – how long the data should be retained in the analytics tools; how frequently you need to do long-term analysis; how urgent it could be; would it be feasible to export data and store it in a data lake so it can be analyzed later? And many more questions like these. Data storage, real-time/near-real-time analysis, faster response adds their weight into the costing factor and hence the retention strategy needs to be carefully looked at.
BI for OTT
The world of video OTT streaming is very different than the e-commerce domain, and requires specific expertise to map the business insights into the analytics data that needs to be captured. It is no longer limited to understanding user demographics (usually captured with cookies and third parties like Bluekai), geography, and payment insights, but is much more than that.
OTT is a complex ecosystem and having an understanding of what to track and why is a very challenging and tricky part. For example, when it comes to video playback, there are many params and events that should be tracked to see different insights like, how much time it took for a video to start loading? Is the delay (potential) with one content or all? Is the user completing the video? How many ads are skipped? What is the exit behavior? And they need params like buffer time, quartile and skip reporting, player event reporting, and a lot many details like title, user level, genre, etc. (the list can go on).
Hence, having the right expertise in video analytics and analysis is very important to devise an analytics plan for a video-centric business.