A feed is a sequence of decisions to keep watching, made roughly every two seconds and mostly unconsciously. Every platform measures those decisions, and every platform distributes based on them. That makes attention the only currency, and the opening line the place where most of it is spent or lost.
This is a writing problem, not a production one. The footage described in how to film better video on your phone can be flawless and still go nowhere if the first sentence gives nobody a reason to stay.
The hook has one job and it is not summarising
A hook does not tell people what the video is about. It creates a reason to watch the next three seconds. Those are different tasks, and the most common opening in business video does the first while believing it is doing the second.
Hi everyone, today I wanted to talk a little bit about our new service is a summary. It asks for patience before offering anything in return. By the time the sentence has finished, the decision has already been made.
Compare: We lost eleven thousand dollars last year because of one line in our booking form. That is not a summary. It is an open loop, and closing it requires staying.
Hook patterns that reliably earn the next three seconds
- The direct contradiction. Everything you have been told about posting times is wrong. It sets up a correction the viewer now wants to hear.
- The specific number. Three things I would fix on your website before spending anything on ads. Specific promises are more credible than vague ones and set an expectation of brevity.
- The named audience. If you run a restaurant and your Tuesdays are dead, this is for you. Narrow openings outperform broad ones because the right person recognises themselves immediately.
- The result first. This email brought in forty two bookings in one weekend. Here is the whole thing. Lead with the outcome and then earn it.
- The mistake confession. I ran ads for two years before I understood this, and it cost me. Admitted error buys attention because it promises something honest.
- The visual cold open. No words at all. Start mid action, with something happening on screen that is unusual enough to hold the frame for a second.
What these share is an implied promise that is paid off later in the video. A hook that does not deliver on its opening claim trains the audience to ignore you, and the platform notices the drop off just as quickly.
Write five hooks and throw away four
The first hook anyone writes is the summary. Writing five in a row forces past it, and the fourth or fifth is usually the one worth using, because the obvious angles have been exhausted.
Test them by reading each aloud and asking what a viewer would lose by scrolling. If the honest answer is nothing, the hook is a description and needs rewriting. It also has to survive being heard without sound, since a significant share of viewing happens muted, which means the first line usually needs to exist as on screen text as well.
A four part structure for everything after the hook
Once attention is earned, the job is holding it. Four sections cover almost every piece of short form business video.
- Hook, one to three seconds. The reason to stay.
- Context, five to ten seconds. Why this matters and who it is for. Long enough to orient, short enough not to lose momentum.
- Payload, thirty to sixty seconds. The actual substance, delivered as two or three points rather than eight. This is the part that gets saved and shared.
- Close, three to five seconds. One instruction. Not three.
The most frequent structural error is a context section that grows into a preamble. If your value only arrives at twenty five seconds, the audience that reaches it is a fraction of the one that started.
Write for the ear, then read it out loud
Text written to be read and text written to be heard are different crafts. Written prose tolerates subordinate clauses, long sentences and formal connectives. Spoken delivery does not, and a script that reads well on a page frequently sounds stilted when performed.
Use short sentences. Use contractions, because nobody says do not in conversation. Prefer concrete nouns to abstractions. Cut every instance of basically, essentially, really and just, which survive editing precisely because they are invisible on the page.
Then read the whole thing aloud at performance pace. Every place you stumble is a place the sentence is wrong, not a place you read it badly. Time it while you are there: most people speak at about 150 words a minute, so a 60 second video is roughly 150 words and nearly every first draft is closer to 250.
Script the visuals in the same document
A script that contains only dialogue guarantees a video of somebody talking at a camera for ninety seconds. Write in two columns, words on one side and what is on screen on the other, and fill both before shooting.
Every point in the payload should have something to look at: a demonstration, a document, a detail shot, a piece of text on screen. This is what the b-roll is for, and deciding it in advance is how you avoid discovering in the edit that no coverage exists for the middle forty seconds.
Write the on screen text as part of the script too. Captions, labels and the opening line as text all belong in the plan, not in the edit as an afterthought.
One call to action, and make it small
Endings collapse under their own ambition. Follow for more, and check the link, and book a call, and share this with someone who needs it is four instructions, which reliably produces zero.
Choose one, matched to the video's job. Educational content earns a follow or a save. Proof and case studies earn an enquiry. Only ask for the enquiry when the video has done enough to justify it, because asking for a booking after ninety seconds of general advice converts poorly and makes the advice look like a pretext.
Make the ask specific and low effort. Comment the word audit and I will send it is a smaller request than visit our website, and it performs better for exactly that reason.
Placement matters as much as wording. An instruction delivered after the value has landed converts better than one bolted onto a video that has already finished saying anything useful, and a closing frame that lingers for three seconds with the ask still on screen gives people time to act on it. Ending abruptly on the last word of the payload wastes the attention you spent a minute earning.
Write the captions, because most people are watching in silence
A large share of feed viewing happens with the sound off, often in public, often as a deliberate default. That makes your on screen text a primary channel rather than an accessibility afterthought, and it changes how the script has to be built.
Auto generated captions are a starting point and consistently wrong about the words that matter most: names, places, prices and anything technical. Those are precisely the words a viewer cannot reconstruct from context, so correcting them is not pedantry.
Keep burned in text short and high contrast, no more than about six words on screen at once, and never place it where the interface will cover it. If your hook depends on being heard, write it as text as well, because otherwise it does not exist for a large part of your audience.
Write once, adapt three times
The same idea rarely works unchanged across platforms, but it almost never needs rewriting from scratch. What changes is the pacing and the framing of the opening, not the substance.
A short form feed wants the payoff early and the whole thing under a minute. A longer form platform tolerates thirty seconds of setup because the viewer arrived deliberately rather than by scrolling. A version embedded on a landing page can afford to be slower still, because the person watching has already chosen your business.
Write the longest version first and cut down, which is far easier than expanding a thin script. The platform differences are covered in our comparison of Reels, TikTok and YouTube Shorts.
Plan the series, not the video
One good video is an event and a format is an asset. Formats compound because the audience learns what to expect, and because writing gets faster once the structure is fixed and only the subject changes.
Look at what you already answer every week. The questions customers ask before buying, the objections that recur, the mistakes you watch people make, the things that go wrong on jobs. Each of those is a repeatable format rather than a single video, and they are already written in the sense that you have delivered them out loud dozens of times.
Batching the writing matters as much as batching the filming. Ten scripts in one sitting is far less work than ten scripts across ten weeks, and our 30 day content plan covers how the two schedules fit together.
Let the numbers edit the next script
Every platform reports where viewers left. That retention graph is the most direct feedback a writer can get and most people never open it.
A cliff in the first three seconds is a hook problem, and the rest of the video is irrelevant until it is fixed. A steady decline through the middle means the payload is too slow or too thin. A drop exactly where the pitch begins means the transition is too abrupt. A flat line to the end means the format works and should be repeated rather than admired.
Read it as instruction rather than judgement. Three videos with the same drop off at the same point is a pattern worth changing; one is noise.
How Social Signals Marketing writes video
We write hooks in batches and discard most of them, script visuals alongside dialogue so the edit is not a rescue operation, and hold every piece to one instruction at the end. Retention graphs decide what gets made again, which is why the formats that work for a client tend to keep working rather than being replaced every quarter.
See our short form video service, read how the platforms compare, or contact Social Signals Marketing about a content programme.
Frequently asked questions
How long should a video hook be?
One to three seconds, which is about five to eight words. The decision to keep watching is made faster than most people believe, and a hook that takes until the fifth second to arrive has already lost the audience it was written for. If it cannot be said in one short sentence, it is a summary rather than a hook.
How many words is a 60 second video script?
About 150, because most people speak at roughly 150 words a minute at a natural pace. Almost every first draft comes in at 200 to 250, which is why reading it aloud with a timer is the most useful editing pass available. Cutting to length nearly always improves the script, because the first things to go are the qualifiers.
Should I script videos word for word or use bullet points?
Script the hook and the close word for word, because those are the two places precision matters most and improvisation is least reliable. The middle can run from bullet points if the speaker knows the material, which usually sounds more natural. The risk with full scripting is a performance that sounds read; the risk with bullets is a payload that wanders.
What makes people stop scrolling?
Something that interrupts the pattern in the first frame: movement, an unexpected image, a face already mid sentence, or text making a specific claim. Openings that look like the beginning of a video, a title card, a logo, someone settling into position and greeting the audience, signal that nothing has started yet and give a clear cue to move on.
How do I know if my script worked?
Read the retention graph rather than the view count. Where viewers left tells you which part failed: an immediate cliff is a hook problem, a gradual slide is a pacing problem, and a sharp drop at the call to action means the ask arrived before it was earned. Look for the same pattern across three or four videos before changing anything, because a single result is mostly noise.