“Using a sledgehammer to crack a nut” – an idiom that should echo in the minds of any video business that is looking to monitor their platform performance. A great solution to any problem requires the most appropriate tool.
Integrating with a video analytics service is the first step, but once you’re all set up, what is the best way to visualize this data in the most effective way? There are almost infinite answers to this question. Every situation and context requires a different approach in order to unlock the right insights into the user experience. This article will take you through the numerous options which we recommend using, what metrics and KPIs best suit each format and what to keep in mind when exploring your data.
It’s a good question. Data is data, it’s pretty black and white. However, this doesn’t mean everyone would choose to portray data in the same way. Lisa Charlotte Rost investigated this exact question in 2016 by putting her data through 24 different data visualization tools and received interesting results…
The most apparent conclusion for Rost was that no two tools are the same. It entirely depends on your goals: “Programming knowledge shouldn’t be a prerequisite to feel home in the data journalism world,” she said. “And let’s face it: Code is still scary for lots of people. We need to work on both sides of the problem. Helping people get into code can make them feel included in the short-term. But we also should work on highly flexible, user-friendly apps which include even more people in the long-term—and in the best case, teach them data visualization rules on the fly.”
Put simply, to visualize your video streaming data correctly, you need to understand which graph, table or chart is going to best portray the trends that help you become data-informed.
A histogram is a type of graph that has wide applications. Histograms provide a visual interpretation of numerical data by indicating the number of data points that lie within a range of values. Basically, this type of graph is used to visualize data frequency using a selection of ranges along the x-axis and the KPI value on the y-axis.
These ranges of values are called classes, buckets or bins. The frequency of the data that falls in each class is depicted by the use of a bar. The higher that the bar is, the greater the frequency of data values in that bin.
At first glance, histograms look very similar to bar graphs. Both graphs employ vertical bars to represent data. The reason that these kinds of graphs are different from bar graphs has to do with the level of measurement of the data. On one hand, bar graphs are used for data at the nominal level of measurement. Bar graphs measure the frequency of categorical data, and the classes for a bar graph are these categories. On the other hand, histograms are used for data that is at least at the ordinal level of measurement. The classes for a histogram are ranges of values.
This is the newest widget we have implemented in YOUBORA Suite and allows our customers to perform A/B tests on variables such as content title to understand the frequency of each metric value for different releases and see a median value to get a quick glance of platform-wide performance.
We like to think of these as ‘intelligent’ tables. Pivot tables are the best way to compare the performance of different variables across a number of metrics. In the example above, we’re looking at a few metrics broken down into five countries: Unites States, United Kingdom, Singapore, Hong Kong and India.
We can clearly see and compare KPI values across dimensions with a pivot table, however, this can then be broken down further. Pivot tables allow you to add multiple layers of filtering in order to compare variables within variables. For example, Let’s say you’re wanting to look at these same five countries, however, you’d also like to break it down into views on mobile phones, TVs and PCs. By adding an extra dimension to this, you can see the total value for each metric for mobile phones, TVs and PCs in all countries.
Pivot tables are a powerful tool, however, its versatility can sometimes make it a confusing way to visualize data. Multi-level filtering allows video services to organize their data into easily understandable segments, however, once you hit three levels of filtering, data can become convoluted. In these moments it is crucial to use a service like YOUBORA Suite which can make these segments make sense using a fluid interface.
A Sankey chart clearly displays a storyline of the events that occur throughout video playback. If you’re wanting to understand what proportion of your users experience each sequence of events, a Sankey chart will break the user experience down into palatable funnels.
In the example above, we can see that Sankey charts allow you to see how many of your plays managed to reach initiation either with or without viewing ads. This allows video services to see where they are experiencing drop-offs and better understand where they should focus their resources in order to maximize ad impressions. If you’re running an AVOD, this format of data visualization is key to your success.
The availability of supply data. Access to real-time data for sequential events is hard to come by as many analytics services do not work in real-time. That’s why using analytics services like YOUBORA Suite is the best option when it comes to working with immediate data. If you are wanting to evaluate risk and value streams, having the right data will make your Sankey diagrams actionable.
Not to be confused with the Sankey diagram detailed above, funnels take much more of a linear approach. This form of observing data allows you to get a complete understanding of where you’re seeing user drop-offs.
In the example shown here, we are looking at how ad rollout is affecting the user experience. The funnel on the left shows one strategy and the funnel on the right shows a second strategy. Using funnels allows you to perform A/B tests capable of revealing how your users prefer to experience your streaming service.
At the heart of funnels is the conversion rate KPI. Before you start filtering by different campaigns, titles and geographies, funnels will not be able to tell you exactly which user segment is affected, only where most users are exiting. Comparing various funnels side-by-side will reveal where your service is strongest, allowing you to replicate this infrastructure in as many situations as possible to make sure all of your users are experiencing the best quality of experience (QoE) no matter where they are or what device they are using to stream content.
It’s easy to get dredged down exploring various graphs, charts and tables in order to find the optimum way of displaying your data. Maps are an often overlooked method of segmenting your audience based on their location and understanding your QoE in various countries, cities and regions.
It could be something as simple as understanding traffic or as complicated as in-stream error ratios. If you’re running a streaming service with a global audience, maps are an unmissable way of visualizing your audience, content and quality data.
Maps are a powerful way of understanding your entire audience at a glance. However, this is also its Achilles heel. Once you have noticed a pattern, you should use more focused visualization methods to dive deeper into your data. However, if you’re starting out your exploration of your platform data, maps are a great place to head first.
Heat maps are useful for visualizing variance across multiple variables to display patterns in correlations. Fractal maps and tree maps both often use a similar system of color-coding to represent the values taken by a variable in a hierarchy. The term is also used to mean its thematic application as a choropleth map.
No matter whether you’re looking at the previous six hours or six months, a heatmap removes any linear elements from your data to simply show your averages. For example, in the image shown, we’re looking at the average buffer ratio during each hour for each day of the week. This data is the average over the course of a month and shows a clear pattern where Saturdays product a lower amount of buffering compared to other days.
Many incorrectly refer to heat maps as Choropleth maps. But choropleth maps include different shading or patterns within geographic boundaries to show the proportion of a variable of interest, whereas the coloration a heat map does not correspond to geographic boundaries.
Because of their reliance on color to communicate values, Heatmaps are a chart better suited to displaying a more generalized view of numerical data, as it’s harder to accurately tell the differences between color shades and to extract specific data points from (unless of course, you include the raw data in the cells).
With Bullet Lists, you can set thresholds for your metrics to better understand how your platform is handling its load compared to your KPIs. Set the minimum, maximum and ideal in-between settings to get a quick shot of how each metric is performing. For example, when looking at Average Buffer Duration, you may want to know whether your platform average is above your maximum setting of 5 seconds(dangerous value), between 2 and 4 seconds(acceptable value) or below 1 second(preferred value).
Bullet Lists allow you to set your own thresholds and understand exactly what your dataset represents. While looking at the average buffer duration in a heatmap may show different intensities of values, Bullet Lists allow you to contextualize your data and understand how many of your users are experiencing a low or high QoE.
Setting your own thresholds is an excellent way of making your data actionable, however, without the proper knowledge of your platform performance, it is hard to determine exactly what a realistic value is for your preferred thresholds. That makes this a fantastic form of data visualization further into your investigation. Use the Map widget or a simple line graph to understand what your lower and higher thresholds stand at and then use a Bullet List graph to distribute your users.
For streaming services, data visualization is the second step. Before this, they need to integrate with a powerful analytics solution like YOUBORA Suite which has the power to group and segment data in exciting ways to reveal new insights into their platform performance. To learn more about YOUBORA Suite, contact our team – email@example.com
Max Gayler on July 11th 2019
It’s that time of year again, fashion designers and models are ready for today’s runway, and NICE PEOPLE AT WORK will be there too! From the 2nd to the 5th of February we will give you live, high quality and uninterrupted streaming of all the shows of the 080 Barcelona Fashion, so you won´t...
YOUBORA Analytics is available in Brightcove’s new advanced player to maximize End User Quality of Experience in OTT delivery Brightcove and NICE PEOPLE AT WORK (NPAW) worked together to enhance Brightcove’s brand-new player with NPAW’s YOUBORA Analytics, the most advanced real-time metrics...
Quality of Experience has become an overused term in the online video industry, but also one of the big challenges that online broadcasters and OTT´s have to deal with. Even more now, a period in which users spend more time connected to the digital world than to reality. Moreover, expectations and requirements...