We recently released our first explainer video, which was about our problem framework. We’d really appreciate any shares:
We created the video as a test. Since videos get more engagement and are seen as best practice in online education, we were considering turning the career guide into a video series. Each article would be 1-2 videos, making for 10-20 in total. This video corresponds to part 4 of the guide. We aimed to make it in the style of Vox’s explainer videos. This is a relatively expensive style, but we thought it could potentially be far more engaging and sharable than our existing articles, while remaining credible.
Although we think the video did well in terms of views and quality, and had lots of positive feedback, we don’t intend to go ahead with producing a video series for now. Below we explain why.
The video has had better reach than an article (figures as of 17 May):
Average cost per view
Note that Google mistakenly gave us about £3,000 of free YouTube advertising, getting us an extra ~85,000 views for free. We excluded this from the “cost per view” figure on the right.
Typically an article might get 10,000-30,000 views in the initial release period if we make an effort.
One interesting finding is how cheap it is to advertise the video on Facebook – just half a cent per view. This makes it looks like we could reach 100,000 people, a material fraction of our target audience, for just $500. Getting a “click” on an article is about 10-20 times as expensive.
However, Facebook views are very short. Now let’s look at engagement time rather than views.
On Facebook, the mean viewing time is only 18 seconds and just 20% of ‘viewers’ turn on sound. So, the 95,000 views are equivalent to about 475 viewing hours. In reality it cost us 40 cents in advertising to get someone to finish watching the video.
On YouTube the picture is quite different. Average viewing time among organic traffic was 4 minutes, and 2.3 minutes among ad traffic. This means we received 1,010 hours of organic viewing time, and about 300 from $280 in advertising spending.
Adding everything, engagement time for the video was about 1800 hours.
In comparison, a good article might get 20,000 views during its initial release with an average reading time of about 2-20 minutes, making for 600 - 2000 hours of engagement time.
As we recently covered, a typical podcast is downloaded by 5,000-10,000 people, and people who start listening using Apple Podcasts on average listen to more than half, likely generating a few thousands hours of engagement.
So, this suggests that videos are better for reach, but about similar in terms of engagement time to a good article and a typical podcast. This suggests using videos more for promotion.
On the other hand, this comparison is unfair on videos. If we built up our channel on YouTube and honed our promotion methods, then we would get perhaps several times as much traffic.
Likewise, if we had a video series, then we might be able to distribute via a platform like Coursera, which might also get us much more traffic.
What’s more, engagement time has limits as a metric. The video conveys some of our most important ideas in just a few minutes, which means the viewer gains much more per minute than they would from a typical podcast. On YouTube, it also has a higher completion rate than a typical article (about 10%).
I was pleased with the quality of the video, but it did cost significantly more to produce than an article. In summary:
60 hours of my time, which is equivalent to 3 weeks where it’s my main focus.
500 hours and $15,700 for the project lead, Maria Gutierrez, a freelancer with a film school background.
$3,500 of other costs, mainly for additional freelancers on sound editing etc.
This was also spread out over about 4 months, which would mean it would take years to produce a series of 10+ videos, unless we could outsource more of the work.
In comparison, writing a new 5,000 word article would take me 1-2 weeks of main focus, and just 1-2 days of extra input from comments and editing. Videos also can’t be edited, which is a major problem since our content is still changing.
The podcast takes Rob 1-3 days, plus another 1-2 days from our podcast producer, and half a day from the guest.
Since the costs of the video are several times as large, we’d need to think the video was at least several times better.
On the other hand, many of these costs were from set-up and learning how to produce videos – and we learned a lot.
If we maintain the style, I’m pretty confident we could reduce the costs by 30% per video, and perhaps 50%.
We could also further reduce the costs significantly by changing the style. Here are some options:
A similar format, but much simpler animation (animation is the main time cost.)
A pure “lecture” style with no animation, as is seen in some Coursera courses.
A conversational rather than scripted video (more like an interview).
Have someone else present them, and perhaps oversee most of the conversion from articles.
However, I expect these changes would mostly make the videos significantly less sharable, reducing traffic several-fold, offsetting many of the cost savings.
All considered, it seems like the potential gains from greater traffic won’t obviously make up for the greater costs.
The most attractive use of a video is promotion of a larger programme. For instance, we could make a 4 minute introduction to 80,000 Hours and explanation of our key advice. We’ll consider making more videos as part of future campaigns.
Changed strategic focus
Another reason to deprioritise the videos is that in the last year our strategy has shifted to a greater focus on “upgrading” existing readers rather than finding new readers, while the video series seems best for finding new readers and conveying introductory content.
We’ve also decided to focus on a more niche and intellectual audience, who we think would prefer podcasts, long form articles and books to a video series.
For all these reasons, we’ve decided to put the video project on hold for now. We plan to revisit it if we move our focus back towards maximising reach and introductory content. We still think having an 80,000 Hours MOOC that’s watched by a significant fraction of college students would be an exciting project to put into action.