What the opening actually has to do
The usual advice is "grab attention", which is useless because it names a goal rather than a job. The job is narrower. In the first second — or the first glance at a static piece — a viewer answers two questions almost without thinking: *what is this*, and *is it for me*. If either answer is missing or ambiguous, they leave, and on a feed leaving costs them nothing.
So the opening is not a trick. It is an answer delivered early. Everything else people call a hook is a way of delivering that answer faster.
On short-form video platforms the drop-off is steepest in the first few seconds — a large share of the audience you lose, you lose before the sentence you were saying finishes. Rather than chase a specific number that changes with every platform update, treat the shape as the fact: attention is not spent evenly, it is spent almost entirely at the front.
Concrete openings that work
- Open on the end state. Show the finished cabinet, then go back to the plank. The viewer now knows what they are being paid for staying.
- Open mid-action. Not "hello everyone, today we are going to" — the hands already moving, the pan already hot.
- Put the question on screen as text. Most feed video autoplays muted. If your opening line is spoken and not written, for a large fraction of viewers it did not happen.
- Make the first frame deliberate. It is a thumbnail whether you chose it or not. A frame of your face mid-blink is a choice you made by not making it.
- For static work, the same rule applies to the largest element. It has to answer "what is this" without the reader taking a step closer.
Captions are the cheapest single improvement available. CapCut generates them automatically and is free on phone and desktop. If you would rather stay in open-source tools, DaVinci Resolve's free tier has automatic subtitle generation, and Kdenlive and Shotcut both accept an SRT file. Whatever you use, check the words — automatic captions are consistently wrong on names, prices and non-English words, which are exactly the words that matter in a local business video.
The ethical edge, stated plainly
Engineering attention is a real skill and pretending otherwise is dishonest — every decision about order and emphasis is attention engineering, including the ones in this course. The line is not between influencing and not influencing. It is narrower and more useful: does the piece pay what the opening promised?
A hook that promises a bamboo house and delivers a bamboo house is craft. A hook that promises a bamboo house and delivers eight minutes of a man talking about his channel is a withdrawal from an account you cannot refill. The mechanism is worth knowing because it is not moral disapproval — it is arithmetic. Bait produces a spike in the first metric and a decline in the second: high clicks, short watch time, unfollows. Platforms have moved steadily toward weighting the second, and audiences remember the last three seconds long after they have forgotten the first.
Three places the edge is clearly crossed, whatever the metrics say:
- A promise the piece does not keep — an arrow pointing at something that is not in the video, a shocked face at a non-event, "you won't believe" resolving into nothing.
- Manufactured fear or urgency — a countdown that does not exist, a health scare, a fake deadline.
- Aiming this at people who cannot easily disengage — children, someone in distress, someone at two in the morning. The techniques work best on exactly the people who are least able to refuse them, and knowing that is part of knowing the craft.
If your opening uses generated footage of something that looks like a real event, label it visibly, in the frame, at the start. Several countries now require exactly that for synthetic media, India's IT Rules and China's labelling rules among them, and the rules are still moving; making things with AI covers where that stands. Beyond the law, a viewer who discovers the opening was fabricated does not carefully separate that from everything else you make.
The check that keeps you honest
Write your opening as a sentence, and then write the sentence the viewer would say afterwards if you delivered. If those two sentences do not match, you have a bait problem, and no amount of retention optimisation is going to fix a broken promise — it will only distribute it more widely.
Do this today
Take your last video and watch only the first two seconds, muted, on a phone, at arm's length, as if it appeared between two other videos. Then ask someone else to do the same and tell you what they think it is about. If they cannot say, the problem is not the algorithm. Fix the first two seconds and republish; it is the cheapest edit in this entire course.
Before you move on