Last updated: 19 August 2026. Written by Adrian Tan, Singapore Digital Marketing (SDM).
Two things kill video scripts, and neither of them is writing quality. The first is arithmetic: someone writes 400 words for a 90-second film, and 400 words takes closer to three minutes to say. The second is process: the script goes to five stakeholders at once, comes back with contradictory notes, and by version 11 nobody remembers what the film was for.
Both are solvable before you write a line. A runtime is a word budget, and it can be calculated. An approval process can be designed so that visuals and words are signed off together instead of in separate, conflicting rounds. Get those two right and the actual writing becomes the easy part.
This guide covers the timing maths, the script formats worth using, structure, writing for sound-off viewing, the Singapore-specific question of register and language, and the approval workflow that stops the version count spiralling. If you are earlier in the process, our guide to planning a corporate video shoot covers what happens either side of the script.
Runtime is a word budget
Start here, because everything downstream depends on it. Narration pace for corporate and explainer work sits in a fairly narrow band. Around 150 words per minute is the conventional baseline for conversational delivery; long-form corporate narration, documentary and e-learning typically run nearer 140; a slower, more considered read for a piece carrying visual information can drop to 120–130. Fast commercial reads push to 170 and beyond, which is why they feel breathless.
The formula is simply word count divided by words per minute. What that gives you is a hard budget to write against.
Three practical consequences follow from that table.
A 90-second company overview is about 200 words. That is roughly one and a half paragraphs of writing. It will not contain your founding date, your five service lines, your values and your awards. The single most useful thing a script writer does is force that conversation early, before anyone has written anything and become attached to it.
Subtract for silence. The budget assumes wall-to-wall talking, which makes for exhausting video. Every deliberate pause, every product shot that needs a beat, every on-screen title costs words. Budget 10–15% of the runtime as breathing room.
Always read it aloud with a stopwatch. Not silently. Reading aloud catches the two things a word count cannot: sentences that are unsayable in one breath, and phrases that are fine on paper and awkward in the mouth. If it runs long — it will — cut a whole idea rather than trimming adjectives across the whole script.
Choosing the script format before you write
“Script” covers four quite different things, and picking the wrong one produces a document nobody can use.
| Format | What you write | Best for | The trap |
|---|---|---|---|
| Voiceover-led | Full narration, word for word, plus visual column | Explainers, product demos, brand films, anything needing precision | Easy to over-write; the visual column gets ignored until the edit |
| Interview-led | No lines at all — a question set and a target story arc | Testimonials, culture films, founder stories, case studies | Teams try to script it anyway, which is the fastest way to make it false |
| Dialogue / scenario | Character lines, action, scene headings | Training, dramatised scenarios, ad concepts | Requires actors and direction; badly performed dialogue is worse than none |
| Text-on-screen only | On-screen copy, timings, motion notes | Short social cuts, sound-off placements, animated explainers | Reading speed, not speaking speed, sets the pace — a different budget |
For interview-led films, the script is a question set, and the discipline is entirely different — our guide to customer testimonial videos in Singapore sets out a working question set and why scripted lines break both credibility and Singapore’s advertising code.
The two-column AV script
Most corporate scripts arrive as a Word document containing only words. That single choice causes most of the revision pain that follows, because it invites everyone to review the copy in isolation from what will be on screen — and then be surprised by the edit.
The two-column audio-visual script fixes it. Left column: what is seen. Right column: what is heard. They are reviewed together, so a stakeholder approving the words is simultaneously approving the visual plan.
| VISUAL | AUDIO |
|---|---|
| Wide, handheld: workshop floor, 6am. Fluorescent lights flicker on. | (no VO — ambient only, 3 sec) |
| Close on hands sorting paper job cards into three piles. | VO: “Every morning started the same way. Forty job cards, three technicians, and a whiteboard.” |
| Cut to whiteboard, half-erased, names in marker. | VO: “By ten o’clock, the whiteboard was already wrong.” |
| Interview: LIM WEI TAT, Operations Manager. Office, window camera-left. | WEI TAT: “We weren’t losing jobs because we were bad at the work. We were losing them because nobody knew who was free.” |
| Screen recording: dispatch view, jobs assigning in real time. Motion graphic: 40 → 3 → 1 screen. | VO: “One screen. Everyone on it. Updated the second something changes.” |
| Logo lockup, URL, 2 sec hold. | VO: “See how it works — book a fifteen-minute walkthrough.” |
Three things this format buys you. Timings become visible, because you can see how much visual coverage each line demands and whether it exists. The b-roll list writes itself — every left-column cell is a shot the crew has to capture. And it prevents the most expensive kind of late note, the stakeholder who approved the words in March and objects to the imagery in May.
Keep the columns genuinely parallel. If a left-column cell says “relevant b-roll”, it is not a script, it is a wish.
Structure: the first five seconds and the shape after them
Structure differs by placement. A 90-second film on your own website is watched by someone who chose to click. A 15-second cut on a social feed has to earn its first second. But the underlying shape of a persuasive corporate film is stable.
The hook is the segment that most repays effort. Twelve words is not much, and the temptation is to spend them on the company name. Spend them on the viewer’s problem instead: the film has to earn its next five seconds before it earns anything else.
On calls to action: one. A film that asks the viewer to visit the site, call the office, download the guide and follow on LinkedIn is asking them to make a decision about decisions, and they will make none. If the film sits on a landing page, the call to action should match the page’s — the logic is covered in our guide to landing page best practices in Singapore.
Writing for sound-off, and writing for subtitles
A large share of video is watched without sound, particularly in social feeds. The most-cited evidence for the benefit of captions is now dated — a 2019 Verizon Media and Publicis Media study of US consumers found people were substantially more likely to finish a video when captions were present — and it should be treated as directional rather than a current Singapore benchmark. The behavioural point holds regardless: if the film will appear in a feed, it has to work muted.
Three script-level implications:
- Write the on-screen text as part of the script, not as an afterthought. If the key claim only exists in the voiceover, a muted viewer never receives it. Put it in the left column as a title.
- Reading speed is not speaking speed. On-screen text needs to be shorter and to hold longer than the equivalent narration. A line that takes two seconds to say takes about three to read comfortably.
- Burned-in captions are a deliverable, not a nice-to-have. Platform auto-captions are inconsistent, and mis-captioned product names look careless. Supply the transcript to the editor. Framing and pacing for feed placements are covered in video for paid social ads in Singapore.
A transcript is also the raw material for the page the video sits on, which is what makes a video findable at all — see video SEO in Singapore for how that works.
Writing for a Singapore audience
This is where imported templates fail. Three specifics matter.
Register. Singapore business English is direct and unornamented. Corporate scripts written in an American register — superlatives, exclamation, “we’re on a mission to revolutionise” — read as insincere to a local audience and, worse, as a company that does not know its market. Write the way a competent Singaporean manager would explain the thing to a peer. Plain, specific, slightly understated.
Singlish. Use it only if it is genuinely your brand’s voice and you can execute it well. A brand that speaks formally in every other channel and suddenly deploys “can lah” in a video reads as pandering. When it fits, it fits beautifully; when it does not, it is the most obvious possible tell that an outsider wrote the script.
Language versions. Singapore’s 2020 Census recorded the language most frequently spoken at home as English for 48.3% of residents, Mandarin 29.9%, Malay 9.2%, Chinese dialects 8.7% and Tamil 2.5%. That is a real argument for language versions in consumer-facing work — and a weak one for most B2B, where the working language is English. If you do produce versions, three rules: budget for translation of the script, not subtitling of the finished film; have a native speaker adapt rather than translate, because idioms and sentence length both shift; and re-time the film, since the same content rarely occupies the same duration in another language. The same logic applies to multilingual web content, which we cover in the corporate video production guide.
The approval workflow that stops version 14
Almost every painful script process shares a structure: the draft is emailed to everyone at once, feedback arrives as tracked changes from six people with different agendas, and the writer becomes an editor of conflicting instructions rather than an author.
Fix it by staging approvals so that each round can only change one class of thing.
- Gate 1 — the one-line premise. Before any script exists: one sentence stating who the film is for, what it must make them think or do, where it will be watched, and how long it is. Approved by the decision-maker alone. Ten minutes of work that prevents weeks of it.
- Gate 2 — the outline. The six segments, one line each, with the word budget shown. This is where structural disagreements belong, and they are cheap here.
- Gate 3 — the two-column draft. Circulated to the named reviewers, with a stated deadline and one instruction: consolidate your team’s comments into a single response. Compliance and legal review happens here, not later.
- Gate 4 — the read-aloud. Read the approved script aloud, on a call, timed. Everyone hears the film. Awkward lines and over-runs surface now, when they cost nothing.
- Gate 5 — lock. The script is frozen. Changes after this point are change requests with a cost and a schedule impact, and they are approved by one person.
Two rules make this work. One consolidated response per stakeholder group, not per person — the client’s job is to resolve internal disagreement before it reaches the writer. And reviewers approve against the Gate 1 premise, not against personal preference: “does this achieve what we agreed the film is for?” is a question that ends arguments, where “do you like it?” is a question that starts them.
The single most common cause of a late-stage rewrite in the Singapore accounts we work on is a stakeholder who was not in Gate 1 appearing at Gate 4. Identify every person with veto power before you write, including the ones who only ever comment at the end.
Where AI helps, and where it does not
Wistia’s 2026 State of Video — based on over 13 million videos and 79 million hours of viewing on its platform plus a survey of 900+ professionals — found that more than a third of teams already use AI in their video workflow, with nearly a quarter more planning to start, and that the primary application is pre-production work such as scripting and planning.
That matches what is actually useful. AI is good at generating a range of structural options quickly, at compressing a long script to a word budget, at producing the 30-second and 15-second cut-downs of an approved 90-second script, and at first-pass translation for a native speaker to adapt. It is poor at the parts that carry the value: the specific detail only your operations manager knows, the sentence that names your customer’s problem in their own words, and judging what a Singapore audience will find overstated. A generated script reads generic for the same reason a scripted testimonial rings false — the specifics are what persuade, and specifics have to come from somewhere real.
A worked example: the 90-second overview
Take the common brief: “a 90-second video explaining what we do.” Here is what the process produces.
Gate 1 premise: “For operations managers at Singapore SMEs with field teams, who currently coordinate jobs on a whiteboard, this film should make them realise their scheduling problem is costing them jobs, and book a walkthrough. It sits on the homepage and as a paid social cut. 90 seconds.”
Budget: 90 seconds at 140 wpm = 210 words, minus 12% for breathing room = roughly 185 words of narration.
Outline: hook on the 6am whiteboard (12 words); the problem, shown not asserted (35); the mechanism, one screen everyone is on (58); proof, the operations manager on camera with a specific number (47); what a week looks like now (35); one CTA, book a 15-minute walkthrough (23).
What gets cut in Gate 2: the founding story, the three other product lines, the awards, and the phrase “industry-leading”. None of them survive contact with 185 words, and the film is better for it.
Notice how much of the work happened before any writing. That is the point. A script is mostly a decision about what to leave out, and 185 words makes the decision for you.
Frequently asked questions
How many words is a 60-second video script?
Roughly 140 words at a standard corporate narration pace of 140 words per minute — about 130 at a slower, visual-led 130 wpm, and about 150 at a conversational 150 wpm. Then subtract 10–15% for pauses, on-screen titles and product shots that need a beat, which puts a comfortable 60-second script at around 120–125 words of actual narration. Always read it aloud against a stopwatch before locking it.
What is a two-column AV script?
A script laid out with visuals in the left column and audio in the right, cell by cell, so that what is seen and what is heard are reviewed together. It matters because it makes timing visible, generates the shot list automatically (every left-column cell is a shot someone has to film), and stops the expensive late note where a stakeholder approved the words months earlier and objects to the imagery at the edit.
Should a corporate video script be written in Singlish?
Only if it is genuinely your brand’s voice across other channels and you can execute it well. A company that writes formally everywhere else and suddenly uses Singlish in a video reads as pandering. The more useful Singapore-specific adjustment is register: local business English is direct and understated, so American-style superlatives and mission language tend to land as insincere.
Do I need to translate my video script into Mandarin or Malay?
It depends on the audience. Singapore’s 2020 Census recorded the language most frequently spoken at home as English for 48.3% of residents, Mandarin 29.9%, Malay 9.2%, Chinese dialects 8.7% and Tamil 2.5% — a real argument for language versions in consumer-facing work, and a weaker one for B2B where the working language is English. If you do produce versions, translate and adapt the script rather than subtitling the finished film, use a native speaker to adapt rather than translate literally, and re-time the edit, since the same content rarely runs to the same duration.
How many revision rounds should a video script take?
Two rounds of substantive feedback is a reasonable target, and it is achievable if approvals are staged: premise, outline, two-column draft, timed read-aloud, then lock. The version count spirals when the full draft goes to everyone at once and the writer is left reconciling contradictory notes. Require one consolidated response per stakeholder group, and have reviewers judge the script against the agreed premise rather than personal preference.
Can I use AI to write a video script?
For parts of it, usefully. Wistia’s 2026 State of Video found more than a third of teams already use AI in their video workflow, mostly in pre-production such as scripting and planning. It is genuinely good at structural options, at cutting an approved script to a word budget, at producing shorter cut-downs, and at first-pass translation for a native speaker to adapt. It is poor at the specific operational detail and the customer’s own phrasing — which is exactly the part that persuades — so treat it as a drafting aid rather than the author.
The short version
Calculate the word budget before you write: runtime times words per minute, minus 10–15% for silence. Choose the format — voiceover-led, interview-led, dialogue or text-only — because they are not interchangeable. Write in two columns so the visuals get approved alongside the words. Spend real effort on the twelve words of the hook and settle on exactly one call to action. Assume it will be watched muted. Write in a Singapore register rather than an imported one. And stage the approvals, because the version count is a process problem, not a writing problem.
Our Singapore client case studies show what this looks like once the films are live and working. If you would rather have the script, shoot and cut-downs handled together, our video production service in Singapore covers scripting through to delivery — and our guide to the types of corporate video is the right starting point if you are not yet sure which film you are writing.
