How to estimate the popularity of events on streaming platforms?

Loading

Organizers, broadcast studios, and creators want as many people as possible to see their content, including broadcasts. The bigger the audience, the greater the reach, the more expensive the advertising integrations, the faster the investments in creating that content will pay off.

How can one estimate the size of the audience? And how to figure out what's more popular in the world: the finals of The International or the Apple presentation? Let's find out how to analyze the popularity of broadcasts. There are several key viewership metrics in the streaming industry, each reflecting some aspect of the event's success: Peak Viewers, Average Viewers and Hours Watched.

Peak Viewers

Peak Viewers is a metric that reflects the maximum number of viewers who concurrently watched the whole event or a part of it: for example, you can calculate the Peak Viewers of a particular match, or the Peak Viewers number at the breaks between the matches. This metric allows to estimate how much attention the audience paid to the key moments of any particular moment in broadcasts: whether it is a decisive moment of the final match at an esports tournament or a presentation by a large tech company announcing its product. 

  Top-5 Esports Tournaments by Peak Viewers (excluding Chinese platforms)  

One should remember that the "peak" of the broadcast can not always be predicted, unless we are talking about a global event (the FIFA World Cup final, the moment of the new iPhone presentation, the Eurovision results announcement, etc). When it comes to esports tournaments, quite often the biggest audience is attracted by the moments when teams with the largest fan bases play. For example, IEM Katowice 2021, one of the biggest CS:GO championships, saw its PV reached at the group stage, while the Grand Final ranked only fourth among most popular matches. That is because the most popular teams of the competition, Natus Vincere and Team Liquid, did not make it to the decisive match. The same goes for various streamer competitions: when one of the famous Influencers is eliminated (and, consequently, usually stops streaming the event), the audience of the event inevitably drops.

Peak Viewers is the most convenient metric for estimating the audience of large-scale but relatively short events. For example, the most popular esports tournament last year was Free Fire World Series Singapore: it was on air for almost 9 hours and viewers spent little time watching its broadcasts in total (compared to world championships in other disciplines), but the concurrent viewership of this event reached an incredible result of 5.4 million people: we should note that in Free Fire and many other mobile games, organizers (as well as title publishers) motivate the audience for such performances with in-game gifts that are given away when certain milestones of peak activity are reached. 

Similarly, it makes no sense to compare UFC fights to long esports tournaments by how much time viewers spent watching them in total. The Peak Viewers number is the most optimal criterion here.

Average Viewers

One of the most popular metrics that many are familiar with is the Average Viewers number. This metric is known by different names. For example, Nielsen agency uses Average Minute Audience (AMA). The streaming platform Twitch previously used Average Concurrent Viewers (ACV or ACCV), and it, just like our company, uses the term Average Viewers.

According to the name, Average Viewers is the average number of concurrent viewers of an event at any given moment of the broadcast. Let's say we are analyzing a major tournament consisting of several stages with different teams in terms of strength and media presence. Obviously, in the first stages (usually qualification) we should expect a smaller number of viewers at the broadcasts, while in the decisive stages of the event (playoffs) the competition begins to reach a much larger audience. The Average Viewers number in this regard better reflects the behavior of the audience throughout the event. At the same time, it should be noted that the events with the same AV number can differ significantly from each other in other indicators.

Let's consider two esports events as examples: Fortnite World Cup 2019 and last year's League of Legends World Championship. Both competitions showed almost the same AV number of between 1.2-1.3 million concurrent viewers, but they differed significantly by broadcast structure and viewership activity at different stages of events. For example, the Fortnite tournament had twice as many streams, and the number of viewers grew from under 1 million on Day 1 to over 2 million on Day 3. In turn, LoL event had the viewership on the streams within the average almost the whole time (the event lasted over a month), and it exceeded the AV number only during the final days. In short, completely different tournaments can show a comparable average audience.

Fortnite World Cup and League of Legends cases clearly show that no single metric is always suitable for determining an event's popularity. The first of these events lasted only three days and featured hundreds of participants, most of whom were not well known to the public, and the final matches predictably caused the biggest hype: thus, it would be more appropriate to determine the popularity of the event based on Peak Viewers number (which was over 2.3 million people). The LoL World Championship, in turn, attracted over 4 million Peak Viewers, but it’s the Average Viewers number that may seem even more impressive, as the competition was held for 30 days and it is worth a lot to reach 1.2 million Average Viewers in such a long time.

There's an opposite example as well. IEM Katowice 2020 (CS:GO) and 2018 Mid-Season Invitational (League of Legends) attracted almost the same number of Peak Viewers at their peak, around 1 million people. But the Average Viewers number at the LoL tournament turned out to be 100K+ higher (382K vs 279K). This indicates that the general interest of viewers in the CS:GO competition was 27% lower than in MSI.

We should also pay attention to the way the Average and Peak Viewers are correlated within the same event. If the PV number is significantly higher than the AV (e.g., by 5-6 times), it indicates that there was only a relatively short period of high viewership within the event. If the difference between both indicators is moderate, it means that the event was interesting for the audience practically all the airtime.

Hours Watched

This metric has several names. For example, YouTube calls it "watch time," Twitch calls it "minutes watched," and Esports Charts uses the term "Hours Watched". This metric is one of the key quantitative indicators in assessing the audience success of any video content, and in simple terms, it reflects the total time viewers spent watching a broadcast/broadcast.

The larger the event itself and the longer its broadcast, the higher the Hours Watched number will be. For example, the World Championships in the most popular computer game, League of Legends, usually last for an entire month and its total airtime exceeds 100 hours: these events are watched by a huge number of viewers, so the total Hours Watched number regularly exceeds 100-150 million. It is not difficult to guess that such events regularly (if not constantly) feature sponsors, who can thus estimate how long their logos and other advertising products are seen by viewers based on this metric.

The Hours Watched metric is calculated using a fairly simple formula: the AV number * Airtime. Thus, if two events have different indicators of the average audience or airtime, then by increasing/decreasing the second indicator, they can achieve the same result in Hours Watched.

  The sum of Hours Watched = the sum of the areas of such rectangles. Accordingly, the more of them (the more points on the segment, i.e., the more often the AV indicator is measured), the more accurate the final data   

For example, let's consider two esports MOBA games tournaments: MPL Indonesia Season 8 (Mobile Legends) and LCK Spring 2022 (League of Legends). Both tournaments were approximately equal by Hours Watched, accumulating 76.9 million and 74.2 million HW, respectively. But they reached this result in different ways: ML:BB event for Indonesia was on air for half as long, but its Average Viewers number was almost twice as high as that of the Korean tournament.

The difference between Peak Viewers and Hours Watched is also worth mentioning. Both metrics are actively used in viewership analytics, and yet they answer very different questions. While Peak Viewers metric indicates the maximum concurrent viewership of an event, the Hours Watched metric gives a general idea of how much time the audience spent on watching the content in general.

What other metrics are important?

In the case of some streaming services, such as Twitch, a popular streaming platform in Europe and North America, we can also take into account additional viewership metrics, which are also extremely useful in assessing viewership success of events.

Video hostings actively use the Views metric, while in streaming it is known as Live Views, and it is calculated differently. One live view is based on the amount of time (a few seconds or minutes depending on a platform) a viewer spent on a broadcast. For example, Facebook counts the number of live views for different time periods: 3 seconds, 15 seconds, and 1 minute.

Average Viewers and Peak Viewers metrics we discussed above capture the number of viewers watching a broadcast/broadcast concurrently. It is also important for companies to understand overall reach: how many viewers watched the event in total? This is exactly what the Unique Viewers and Unique Authorized Viewers metrics reflect. The first captures the total number of people who watched a unit of content, and the second captures the number of authorized users of the broadcast service (Twitch) who joined a stream.

At first sight it may seem that the first of these two metrics is more universal and covers a larger number of people. This is true, but at the same time it will not help in the analysis of audience engagement. The behavior of Unique Authorized Viewers can be tracked, and thus reach, engagement can be estimated based on this. For example, an event broadcast may have various promotional integrations designed specifically for the authorized audience (e.g., prize give-aways among chat users), and the event sponsor wants to understand the audience that writes in the chat and the reach of such integration: the Unique Authorized Viewers metric is exactly helpful in assessing this.

When you know the Unique Authorized Viewers metric, you can also count more specific metrics that give a broader picture of viewership engagement in broadcasts. In particular, you can find out the average time such users spent watching the event: whether viewers stayed with you for most of the broadcast or whether they joined your stream and almost immediately switched their attention to other broadcasts.

Let's look at examples of two Twitch streams that are comparable in terms of basic viewership metrics, but quite different by additional metrics. We can compare the Minecraft Twitch Rivals finale by the Spanish streamer auronplay and the CS:GO Major final on the PGL channel. Both broadcasts showed almost the same Peak and Average Viewers results and even were almost identical by the number of Unique Viewers. But the viewers’ engagement was different.

  Average Viewers number o PGL was significantly lower in the first half of the broadcast than in the second one. In turn, Auronplay has a more smooth chart 

For example, the PGL stream attracted many times fewer live views, largely due to the fact that the first third of the stream was dedicated to the showmatch, which was watched by significantly fewer people than the competition final itself, while the auronplay's stream was much more "even" by Average Viewers. Interestingly, the PGL broadcast was watched by far fewer Twitch Unique Authorized Viewers than the Spanish influencer's stream, which can be attributed to the fact that many people watched the PGL broadcast not on Twitch itself (viewers could watch the match via the embedded players on other sites).

In turn, the average viewing time per Unique Viewer on the PGL broadcast was higher than on the auronplay's one: this is because of the features of the competitions broadcasts. The Twitch Rivals event was broadcast by dozens of participants, and the Spanish audience regularly switched between broadcasts to watch the competition from the perspective of different creators. The CS:GO tournament final is not appropriate for that.

To sum up, additional metrics give a more in-depth understanding of the broadcasts popularity and help answer more specific questions: how many people in general were attracted by a broadcast, how intently a viewer watched a broadcast, whether people watched a stream directly on his page on the streaming platform or used third-party sites, where the broadcast player was built-in, etc.

***

Esports Charts has been collecting and analyzing statistics on both esports tournaments and streamer events, as well as other global events that are streamed on major services such as Twitch, YouTube Live, TikTok, Facebook Gaming and a lot of other platforms for more than six years. If you want to find out about the popularity of your live-streaming project, contact us via [email protected] — and we will reply to you as soon as possible.