For video services, there’s nothing more damaging to your return on investment (ROI) than going in blind. For many, the cost of real-time analytics doesn’t seem worth the trouble, but when a service performs poorly and isn’t able to engage with its audience, it’s often problems that could have been avoided if only they had been noticed immediately.
It’s not only the video industry that is becoming obsessed with ‘real-time’. The mobile app industry, FinTech and even the transportation industry are looking into ways to optimize their pipeline and reduce room for error. So, why is there such an obsession with real-time analytics? This is something this blog is going to discuss in length and break down into the various verticals which can be supported with complete transparency of video data.
Online video and Pay TV are complex industries with technical delivery paths which require skilled data scientists to interpret various errors and anomalies within the content journey. However, video analytics is able to translate all of this information into easily-digestible KPIs which can be investigated using a third-party software, or in the case of YOUBORA Suite, through fully customizable dashboards.
According to Gartner, real-time analytics means that the analysis happens within seconds after new data is received. Gartner differentiates between two kinds of real-time analytics:
- On-demand real-time analytics waits for a user to request a query and only then provides the analytics results.
- Continuous real-time analytics analyzes data continuously and sends alerts to users whenever certain events happen.
Gartner’s definition helps us break down real-time analytics into two channels: Live and VOD.
If you’re a broadcaster or streaming service that works with live streaming, the essence of ‘real-time’ is inherently present. When you are working with video content that is being experienced simultaneously amongst thousands or even millions of devices, it is crucial that any platform anomalies are detected immediately. Without the proper infrastructure, your video service is going to be facing many quality problems from overloaded CDNs, unpredictable traffic and unexpected crashes that may stem from crowded servers.
For instance, if we take a look at the Australian streaming company Optus during the World Cup 2018. Their streams were so overloaded that the video content became unavailable for everyone, ultimately marring the brand and leading to lost impressions for the company. These problems could have been avoided with the proper infrastructure and real-time analytics which could have alerted them to platform issues the second they occurred, allowing Optus’s operational and technical teams to fix the problem or at least allow customer service teams to contact users and calm the fire.
You can learn more about how to stream global events like the FIFA World Cup 2018 in our whitepaper, ‘The 2018 World Cup in Data’.
While a different beast altogether, real-time analytics remains a failsafe option for streaming services looking to stay afloat in the competitive OTT landscape. While this technology may not serve in relation to Live content, it is important that users of SVODs, AVODs and TVODs are served the best quality of experience (QoE) possible at all times.
There are numerous KPIs which should be tracked in real-time. Not only does this keep your operational teams up-to-date to the nearest second of any platform errors, but it also allows you to track how your customers behave on a granular level. This means tracking page traffic, concurrent plays, buffering or platform errors in real-time. If you notice an upward trend in any quality metric, your operational teams are able to act immediately and work with actionable data.
Benefits of real-time
Respond to network, CDN and regional issues immediately.
If something goes wrong on your platform, you won’t have a lot of time to fix it before it starts causing real damage. When there’s a sudden change in the key metrics that you’re tracking, you need to know immediately.
Real-time analytics isn’t just about monitoring your entire platform. For many streaming services in 2019, a global audience is consuming content and has to be monitored simultaneously. By segmenting your audience, you are able to discover struggling areas and take proactive measures to optimize your platform.
Let’s say you’re an AVOD with users in more than 20 countries. It’s easy for any OTT company or broadcaster to look at your video service and say, “There’s a lot of buffering on this video”, but with real-time monitoring, businesses can be more confident in problem identification. For example, “I can see that in the past two hours, all my users in North America showing mid-roll advertisements experienced an average of 10% buffer ratio which is much higher than usual.” With analytics, video services can zoom into their issues and immediately tackle platform problems before users start to churn from poor QoE.
Multi-dimensional filtering is a capability of YOUBORA Suite that allows streaming services to infinitely break down their data until the struggling areas are uncovered. You may notice that your platform average time between rendition switches is shrinking, signifying that players are struggling to load HD quality video. You can then filter by country or any dimension to uncover whether this problem is coming from a specific country, device or CDN. This can then be broken down further to see if it is a combination of factors affecting player rendition switches. Operational teams can then take action with this data and focus on fixing this issue.
This blog has spoken a lot about ‘monitoring’ and ‘tracking’, however, this doesn’t insinuate someone staring at dashboards 24/7 in order to remain as ‘real-time’ as possible. Many analytics solutions such as YOUBORA Suite include an alerts feature to avoid this. This will use either preconfigured thresholds or an intelligent algorithm that will use historical data to understand what counts as dangerous metric values.
This means that with real-time analytics implemented and real-time alerts, streaming services are able to focus their resources on other areas. This is ultimately a money-saving investment and allows the right departments to be notified of worrying platform errors on the right channels.
The implementation of a Real-Time Big Data Analytics tools may be expensive, but it will eventually save a lot of money. There is no waiting time for business leaders and in-memory databases (useful for real-time analytics) also reduce the burden on a company’s overall IT landscape, freeing up resources previously devoted to responding to requests for reports.
By observing platform trends, you are able to better understand where resources need to be focussed and use these patterns to predict the future for your service with actionable data. It could be that you are spending a lot of money on the infrastructure of your users in France, however, with analytics, you could find that France does not require as much bandwidth as previously thought. Where the ‘real-time’ element comes into this is load balancing. If you’re running a multi-CDN content path, it is important that you configure your CDNs correctly. With the use of CDN Balancer technology, you can automatically use the right CDN for each user in real-time.
Simply, analytics helps you make good decisions fast. Videos aren’t predictable once you release them into the real world. Since you can’t know the future, you should at least know the present — and that’s why real-time analytics, that self-updates all the time, can help. These tools provide you not just with real-time data, but also with automatic, instant analysis of that data for whatever purposes you need, from attribution sources to conversion funnels. If your dashboards show the most up-to-date data, they’ll give you the power to make good decisions fast. You won’t have to resort to guesswork when you’re under stress.
Operational teams aren’t the only ones who can benefit from real-time analytics. Product teams can use data to understand how different content performs. Performing A/B tests with different website layouts or changing your recommendation algorithm can be monitored as it happens so you’re not having to waste any time in improving your content catalogue.
Be data-informed to stay ahead of the competition. – Having a robust analytics solution built into your player can help those responsible for buying programming make more informed decisions based on the tastes of their audience. Robert Farazin of TVBeat has said that: “Content spend is usually the highest expenditure within a broadcaster, representing around 40% of overall budgets. There is so much content nowadays that you cannot buy everything, so you really need to optimise for your users. If you don’t have analytics, then you don’t know what your subscribers are watching.”
Your users are complicated, and constantly behaving in weird, unexpected ways. The only way to catch up with them is to know what they’re doing on your platform right now. That’s something only real-time analytics can provide, by constantly collecting user data and crunching the numbers to give you whatever insights you need.
Best KPIs to track
Data Management System (DMS) with Business Analytics concept. businessman working with provide information for Key Performance Indicators (KPI) and marketing analysis onn virtual computer[/caption]
No matter what real-time analytics platform you’re using, it’s important that you’re tracking the right metrics. We have put together a list of the most crucial metrics to track when it comes to real-time tracking.
Buffering is what happens when a video player fails to pre-load a sufficient amount of content data and the playhead stop so that the player has time to load more content. This is often associated with the infamous ‘buffering wheel’. The Buffer Ratio is the percentage of a users video session they spend experiencing buffering. The ratio is obtained by dividing the buffering time by the total playtime. When calculating the ratio, initial stream-join buffering is excluded. So, should your Buffer Ratio be high or low? As low as possible.
The higher the buffer ratio, the higher the proportion of time users spend waiting for a video to load. For example, if you’re noticing that the average buffer ratio of your platform is anything above 5%, your users are spending way too much time staring at the buffering wheel.
Simply, the number of concurrent plays the total ongoing views at any time. This metric will show you how many users are viewing content in real-time. While this may be perceived as a vanity metric, when we start to filter data, this metric shows us where most of our traffic is coming from, what device is being used by most of your viewers or even which title is the most popular. This metric allows you to focus your efforts on the areas with the most traffic and segment your audience.
While being aware of the number of concurrent views is paramount, it’s equally as important to know how many views are being abandoned. There are different benchmarks for every business but it’s important to monitor your concurrent plays and know immediately if you’re starting to lose viewers.
In-stream Error Crashes
If you’re tracking the number of concurrent plays and notice it dropping, this should be the next metric you check. It could be that users are dropping because their content is crashing. These errors need to be acted upon quickly in order to not leave a bad taste in users’ mouths.
Like with all metric, the benchmark number of views crashing depends on the service, however, it is important that this is monitored.
Real-time analytics is no longer an exclusive add-on for the big players in video streaming. No matter what size your business is, the cost-saving and platform boosting elements of real-time analytics can make the biggest changes to even the smallest of business. To learn more about YOUBORA Suite, go here.