Visualised #3: National Theatre at Home
Since cultural venues around the world were forced to shut their doors a few weeks ago, there’s been an explosion in online streaming of theatre, dance, opera, music, and more.
One such is National Theatre at Home, with weekly broadcasts of plays from the NT Live catalogue every Thursday evening. The shows are free to watch, although a donation is encouraged, and they stay online for a week. The first to be shown was the hit comedy, One Man, Two Guvnors.
The National Theatre have been using YouTube for their broadcasts and also YouTube Giving to encourage donations. This was interesting to us because:
YouTube shows a count of concurrent viewers during live broadcasts, and then total video views afterwards
The donation box also shows a live count of the amount raised
Live chat was turned on, allowing viewers to respond and interact
So what we had was a popular livestream event, with some potentially interesting data up for grabs. We collected what data we could and have done some quick analysis. Here’s what we found.
Viewers
We looked at the number of people who watched the live broadcast. The chart below is also marked up with the beginnings of acts 1 and 2, the interval, curtain call, and credits.
There were about 120k people ready and waiting when the curtain went up. Our data starts 15 minutes before the scheduled start, by which time 50k people were already there.
The number of viewers peaked at around 216k and stayed remarkably consistent throughout.
There’s an interesting little wobble around the interval, but numbers came back up. The dip just after 7pm (when the show started) looks like it was a glitch at YouTube’s end.
As for the dip around 8pm, that will have been for the Thursday night clap for the NHS (with thanks to Ed Vaizey and others for pointing that out).
It’s worth mentioning that although the metric here is ‘viewers’ it’s only really counting the number of devices it was streamed to.
A week later, just before the show was taken offline, the total number of views stood at 2.6 million. That means around 8% of views came during the live premiere, and 92% came over the following 7 days. That’s interesting in the context of our next subject.
Donations
As well as the number of viewers(/devices) viewing, we also recorded the cumulative count of donations made via TouTube.
$41,600 was raised by the end of the premiere. That’s about $0.2 per viewer. Of that…
$16k was raised before the start
$8k was raised during Act 1
$3k was raised during the interval
$3k was raised during Act 2
$11k was raised after the curtain came down
When the page loads on YouTube a donation request is shown prominently to the user. Still, I was surprised to see how much was given before the show even started.
I was also surprised by how little was given during the interval, when a big call to action was displayed, although this was a useful reminder that paid off when the show ended.
A total of $82,500 shortly before the stream was taken offline a week later. If 2.4m views came later, then that’s about $0.02 from each of those viewers.
That also means that the 8% of viewers who watched the show live were responsible for around 50% of donations.
Comments
We captured 7,885 comments from the live chat. There were another 1,900 left underneath the video during the following week, but we don’t have those.
The chart below shows the number of comments left per minute before and during the broadcast.
As you can see, there’s lots of activity and excitement before the curtain goes up, and things trail off through the performance.
That said, there’s still a fair bit of activity going on deep into Act 2, with new comments every few seconds. The interval goes very quiet though.
It’s also interesting to see the relatively sustained spike just before the interval. Act 1 ends with a superb bit of farce, and that’s what people are reacting to.
A few tidbits:
There were 2,382 different commenters
The National Theatre’s account commented 112 times
977 comments included emojis
208 comments mentioned keywords relating to school, teachers, or similar
We used a text analysis tool called MonkeyLearn to assess the sentiment of the comments. Here’s how that looked.
It’s worth mentioning that very few of the negative comments were made in reference to the production.
We’ve got a bit more to add on the text analysis side and will try to update this later.
Some other observations
This has been one of the more popular livestreamed productions of the past few weeks and was undoubtedly a success from the point of view of contributing the NT’s mission to show “world class theatre that entertains, inspires and challenges the broadest possible audience”.
Different organisations and trying different release schedules for their broadcasts. I think the NT at Home model of weekly live broadcasts followed by a week to catch up is a good one. That gives them:
the benefits of an ‘event’ with time to build up a concentrated message, a communal moment, and a bump in donations
the benefits of catch-up, where you seem to get the majority of views and a few more donations
the chance to form a habit with a regular audience
Compare that with the Met Opera’s approach of a different opera every night which makes it harder to make the most of every show (from a marketing and education/interpretation POV) and risks exhausting the most devoted audience.
Finally, on the financials, $82k isn’t to be sniffed at but it’s worth bearing in mind that:
The cost of filming the production in the first place would have been considerable
The 2019 NT Live encore screening of One Man, Two Guvnors grossed £1.2m in the UK and Ireland alone (data from the Event Cinema Association)
Which makes you wonder what the business model for purely digital distribution of this sort of thing might be.
Behind the scenes
For a minute-by-minute count of viewers and donations, I used a workflow in Automator to take screenshots every 60 seconds and then did some manual data entry.
To capture the live chat comments I used the Save Live Streaming Chats for YouTube Chrome extension.
The text analysis was done using queries in Google Sheets. We also leaned heavily on MonkeyLearn which allowed us to use machine learning models to extract topics, keywords, sentiment, and abuse/profanity (with huge thanks to the very helpful team for their support).
For the YouTube Subscriber data I used Social Blade.
Want more visualisations?
This post is the third in our Visualised series where we take cultural sector data and make it more accessible through visualisations.
Sign up to our newsletter and we’ll tell you when new ones are published.