🔎 Retention Review #7 - He fixed retention AFTER uploading
Dissecting your retention graphs to figure out why people stopped watching.
Hey everyone,
Welcome back!
It’s the first retention review of 2023! We have some awesome case studies this week.
We’re talking about the importance of video vs audio, the problem with fast cutting, and how to fix retention after uploading.
And stick around til the end for some exciting news 😏
1 - Sound vs Video
Title: Bell's Theorem Proves That Our World is Quantum, it Cannot Be Mechanical
Creator: PhysicsLaure
Average % viewed: 38.8%
This video breaks down Bell’s Theorem in a way that even I could understand (mostly 😅)
What to repeat:
Style
Breaking down science into digestible content is hard and all her videos do this well!
What to improve:
Audio
When I was working for Ali, I remember it was drummed in over and over that getting a good mic is way more important than getting a good camera. This video proves that.
The blurry audio is immediately noticeable and puts me off, sadly.
That probably explains the 40%+ dropoff in S1.
Transitions
The varied visuals are good, but the transitions between scenes sometimes feel a bit strange.
Rather than smoothly transitioning from visual to visual, we fade to black almost every time, even mid-sentence occasionally. This affects the momentum.
It would feel more natural to fade to black less frequently (maybe at the end of a segment) rather than at the end of a sentence.
Take this recent example from Kurzgesagt:
Obviously, the visuals are absolutely world-class and incredibly difficult to emulate, but notice how natural the visual transitions feel.
Rather than fading to black, they flow smoothly from one to the next.
The information is delivered at a similar speed to Laure’s video, but the momentum feels more consistent because of the visuals.
2 - Cut, cut, cut…
Title: I Made $___ Selling AI Generated Art
Creator: Thomas Chow
Average % viewed: 42.5%
Thomas generates AI art and tries to sell it. But how much can he make in a month?
What to repeat:
Concept
The video is fun.
I liked the different approaches to making money, first with Fiverr, then by printing the art and selling it physically.
Hook
Great opening 10s.
It’s intriguing, there’s humour, and it provides great context for the video.
Yes, there’s an overall steep drop over the first 40s, but this could be explained by what I’m about to discuss next…
What to improve:
Pacing
The fast cutting feels overwhelming after a while.
It isn’t terrible by any means, but it feels like the aim is to just cut, cut, cut to keep the momentum going - even cutting within sentences which gives the speech an unnatural feel at times.
The fast cutting works well in the hook, but I would have liked to see things slow down for a second once the video properly started.
Take this recent example from Ryan Trahan:
The hook is pacey (and timed to music which is really fun), but as soon as the hook finishes, you’ll notice the pace slows way down.
The effect this has is that we feel excited and intrigued during the hook, but then we feel like we’re actually getting to know Ryan when the pace slows.
It also causes pattern interrupt which re-engages our minds.
Whereas in Thomas’s video, the non-stop cutting made me feel like we were rushing to get through the content.
At the start of S3 we’re still racing through the information. This may have contributed to the dropoff.
3 - Edit after upload…
Title: I Bought HDZero Goggles, This Is What They're Like
Creator: Darren Allatt
Average % viewed: 49.3%
Darren’s review of these HDZero goggles had a bit of an early dropoff… so he took action to fix it!
What to repeat:
Additional cuts
You’ll notice some classic early dropoff in S1, but Darren wasn’t satisfied with that.
He realised he’d spent so much time giving context, he wasn’t talking about what people really cared about (the review) until 0:38.
He’d given a full rundown of everything that would be covered in the video, as well as taking an additional 10s to introduce himself and explain what to expect from his channel.
Important info? Yes and no. But definitely not worth spending 27s on.
Through the YouTube editor, he cut out this whole segment, and the review now begins at 0:11.
And from that point, retention is mostly completely flat. There’s great switching between A-Roll and B-Roll, shots of the product, footage of the drone flight, etc. It’s a great review.
Lesson to be learned here - it’s not too late to make changes once the video is uploaded.
It may not always be possible, but it’s worth checking your retention after a few days and figuring out if you can cut anything that doesn’t serve the video overall.
What to improve:
CTA
Darren asks twice in this video for people to comment, but doesn’t direct them to another video at the end.
This comes down to personal choice, but I would usually recommend pointing people to the next video you’d like them to watch.
Choose a video that covers a similar topic, or which reveals information the viewer feels like they want (but still don’t have) after watching your current video.
This increases the likelihood the audience will keep watching your content, increases their session time, and indicates to the algorithm that your videos are worth pushing to more viewers (because they keep people on the platform for longer).
That’s all for this week!
But there’s one last thing…
📈 Retention Graph Database
I’m launching this brand-new database in the next few weeks.
Every retention graph I’ve ever reviewed, all in one place.
Search by niche, by AVD%, by views, by video length, or even just by scrolling through the graphs themselves!
There will be discounts for you (as a loyal reader!) PLUS even bigger discounts for anyone who’s ever contributed a graph.
More info soon 😁
See you next week!
George 👋