So, I was thinking about how content creation is integral to my growth plan, both for myself and for the softwares I’m going to create. I cannot just have it very wishy-washy when it comes to how I plan to grow using content or how I create content.
My approach to it is to see a video as an amalgamation of video clips and sentences. What I mean by that is each script on reels for me is like 7-8 different sentences that I have put together. The nature and the sequence of sentences matters to me, and those sentences are represented visually using video clips. It can be Aroll of when I was speaking those sentences, or it can be B roll.
My eye is keenly looking at the results that are produced, and looking at how the sentences in it were structured or laid down in a sequence. Same for the clips also.
For now, my research has shown me that for my page, the best results I get are when at the start, the video is an A-roll clip of my face close-up and then a B-roll clip of me working on my computer, and the sentences are a hook and then my introduction. But this is not the optimum, because the skip rate does not reliably stay below 35% all the time.
But I can not even get comfortable with it.
Even if I find an angle where I can comfortably always get 35% or below skip rate, a change in the platform can erode that characteristic. So, what I’m actually after is 4-5 different formats or skeletons that hold up 35% skip rate and are always in search of skeletons that can get me 35% or low, so that I can always adapt. What I’m essentially looking after is not a creative way of writing scripts but a system which allows me to create permutations and combinations of sentences and video clips which can always allow me to adapt and end up with 35% or less skip rate.
And what does that system look like? I think I need something that will give me my transcripts and the performance so that I can judge everything on a sentence level. Everything is about the sentences and also mapping out sentences to a video. My current rule is that a sentence or two sentences together shall be represented by only one clip. Like I do not allow myself to have more than one clip per sentence. Basically, two videos won’t represent a single sentence is what my philosophy is right now. Just for data’s sake also.
I need to be able to connect videos and sentences and analyse their performance as per sequence. I need to create a software for that. Content creation is way more systemic and iteration-based than creativity. Stay beholden to creativity, I will burn out and won’t get anything I want. I need to systemise and run permutations and combinations of things that matter so that I can always stay in the Goldilocks zone.
I want to extensively study the first 5 seconds of each of my videos and the skip rate because it is the Pareto generator. I believe if you have a good skip rate, even by 5%, you can increase your views by 10x.
Example data can look like this:
Here’s a demonstration table focusing on the critical first 5 seconds:
Video Performance Analysis – First 5 Seconds
| Video ID | Sentence 1 | Clip 1 | Sentence 2 | Clip 2 | Sentence 3 | Clip 3 | Overall Skip Rate |
|---|---|---|---|---|---|---|---|
| V001 | “I spent $47K building this app” | A-roll (face close-up) | “and nobody used it” | B-roll (computer work) | “Here’s what I learned” | B-roll (app interface) | 31% |
| V002 | “Most founders quit too early” | B-roll (startup office) | “I’m building in public” | A-roll (face close-up) | “and sharing everything” | B-roll (computer work) | 45% |
| V003 | “This coding mistake cost me 6 months” | A-roll (face close-up) | “Let me show you” | A-roll (face close-up) | “so you don’t repeat it” | B-roll (code screen) | 29% |
| V004 | “Why your SaaS isn’t growing” | Text overlay on B-roll | “I figured it out the hard way” | A-roll (face close-up) | “Here’s the system” | B-roll (computer work) | 40% |
| V005 | “I analyzed 50 viral reels” | A-roll (face close-up) | “to find this pattern” | B-roll (data/charts) | “that nobody talks about” | B-roll (computer work) | 33% |
Pattern Insights from This Data:
Winning Formula (Sub-30% 3-sec skip rate):
- V001: Hook (personal stakes + number) → A-roll face → Conflict/surprise → B-roll
- V003: Hook (mistake + time cost) → A-roll face → Promise → A-roll face
- V005: Hook (research + number) → A-roll face → Value tease → B-roll
Losing Formula (Above 35%):
- V002: Started with B-roll instead of face
- V004: Text overlay on B-roll doesn’t grab attention
Key Finding: Your hypothesis is validated – A-roll face close-up in position 1 (Sentence 1/Clip 1) with a high-stakes hook performs best. Position 2 can flex between A-roll continuation or B-roll transition.