Use ChatGPT to turn product facts into three English short-video scripts

Turn product facts into three English openings, a four-shot filming plan, timed narration and an editing handoff, with complete prompts and revision examples.

The goal is specific: three English product-video scripts for a fifteen-second format. All three cover the same confirmed point, changing only the opening and narrative order so you can choose one for filming or a video tool.

ChatGPT organizes information and drafts the copy here. Product accuracy and conversion performance still require production checks and real testing.

Prepare a fact sheet and matching original photos

Prepare the product name, image IDs, confirmed specifications and sources, target market, audience and one content goal. No API connection is required for these prompts. Attachment features depend on your current account interface.

Start a new ChatGPT conversation and paste the fact sheet. If using images, identify them as references for appearance and composition, not proof of unspecified materials or performance. You can start with complete text information when images are unavailable.

Round one: check the facts before drafting an ad

Act as my overseas product-content editor. Check the material first; do not immediately write promotional copy.

[PRODUCT FACTS]
[Paste the product-fact brief, including sources and image references.]

[CONTENT TASK]
Market: [country or region]
Audience: [specific people and use setting]
Language: [for example, British English]
Aspect ratio: 9:16
Target duration: 15 seconds
Core message: [one point only]
Viewer action: [choose one: save, comment, view details or enquire]

Return three columns:
1. Confirmed facts we can use directly.
2. Ideas that must remain creative hypotheses.
3. Questions you must ask me.
Do not add unspecified performance, certifications, customer experiences or sales figures. Ask before proceeding if a missing answer affects product facts.

Look for assumptions such as turning “looks metallic” into “stainless steel,” or “separate compartments” into “saves half the space.” Correct the fact sheet before moving on. Retain the units, conditions and sources of real measurements.

Round two: one selling point, three openings

Write three English short-video scripts using the facts we just confirmed.

Version A: open with a specific situational problem.
Version B: open with a product-use action.
Version C: show the finished state first, then explain the process.

Change only the opening and narrative order. Keep product facts, audience, core message and closing action consistent.
Start with 25–35 spoken English words per version and include the Chinese meaning. This word count is a drafting starting point, not a guarantee that it will fit; state that a recorded timing check is required.
Use short, natural sentences. Avoid exaggerated effects and invented customer testimonials.
Return the version name, what the first three seconds should communicate, full voiceover, closing action and the evidence required for product claims.

Script examples: three ways to describe the same tray

This exercise uses a fictional white cutlery tray with three fixed compartments. The only confirmed action is placing spoons, forks and knives in separate sections. Time savings, space changes and fit have not been tested. These are editorial examples, not advertising results.

A · Situational problem
“Everything in one pile? Give spoons, forks, and knives their own section. Here’s one way to arrange your cutlery. Save this idea for your next drawer tidy.”

B · Action first
“Spoons here. Forks here. Knives here. Three sections, one cutlery layout. Watch where each piece goes, then save this setup as a reference for your own drawer.”

C · Result first
“Here’s the finished layout: spoons, forks, and knives in separate sections. This is how we placed them, one group at a time. Save the idea for later.”

Version A starts with the mixed-cutlery problem, B leads with sorting, and C shows the finished arrangement. None claims greater speed, more space or universal drawer compatibility. Choose one version before production so the objective does not keep changing.

Round three: turn the selected copy into executable shots

Use version [A/B/C]. Create a 15-second shot table:
Time range | Subject | Exact action | Product state | English caption | Matching voiceover | Reference-image ID | Acceptance criteria.

Keep these conditions fixed:
- The same tray, cutlery set, drawer, camera position and lighting.
- The number of spoons, forks and knives must stay the same.
- Specify where every action begins, its movement and its ending position.
- No hands passing through the product, changing structure or disappearing objects.
- Do not introduce any unconfirmed product effects.

Start with four shots and identify footage that must be filmed. Produce only a script and shot plan; do not imply filming, testing or publishing has already happened.

Before handoff: check facts and time a read-through

Compare every line of English against the fact sheet and check that each shot has usable materials. Read the script at a realistic voiceover pace and record its duration. If it runs long, shorten the sentences rather than rushing without pauses.

Keep four deliverables: the fact sheet, selected script, shot list and missing-materials checklist. In the first test, change one main variable, such as the opening. Record saves, clicks and enquiries separately instead of treating views as sales.

Round four: turn version B into four shots

I would develop version B first because it opens with an action. Align one video before producing all three alternatives. The following timings are a filming and editing plan and must be adjusted to the actual narration.

Shot 1, 0–3 seconds: fixed overhead view, a right hand puts spoons in the left compartment; narration “Spoons here.”; caption “Spoons”. Begin before the hand obscures the tray. Shot 2, 3–6 seconds: place forks in the middle compartment from the same position; “Forks here.” and “Forks”. Shot 3, 6–9 seconds: place knives in the right compartment; “Knives here.” and “Knives”.

Shot 4, 9–15 seconds: remove the hand and show all three compartments. Narration: “Three sections. One cutlery layout. Save this idea for later.” Show “Three sections” and “Save this idea” in sequence. This shortens the earlier B script and removes the repeated instruction to watch each placement.

Capture stable handles before and after each action for editing, keeping utensil counts consistent. Name usable clips B01–B04 and reshoot missing actions. A photo filename is not a completed video clip.

Using the confirmed white three-compartment tray, prepare a filming and editing handoff for version B.
Final narration: Spoons here. Forks here. Knives here. Three sections. One cutlery layout. Save this idea for later.
Shots: B01 spoons; B02 forks; B03 knives; B04 complete layout after the hand exits. Keep product, lighting, overhead angle and utensil counts consistent.
Target 15 seconds, 9:16. Timing is planned, not measured. Without footage, do not claim to have inspected it.
Deliver a four-row table with shot ID, planned interval, starting state, action and ending state, narration, short English caption, required asset and review point; separate clean narration without notes or shot IDs; separate caption copy without fabricated precise SRT timing; the actual recording measurements needed to revise timing; and three likely failures with specific revision instructions.
Add no product features. Music suggestions, automated publishing and advertising plans are outside this task.

Round five: revise against the actual recording

Record the script in the voice you intend to use. Note its total duration, sentence starts, pauses and awkward pronunciation. If it exceeds 15 seconds, remove repeated ideas and long modifiers before increasing speed. If it is shorter, leave time for actions and the final view rather than adding claims.

Give the recording to the editor and align captions to actual speech. Text alone supports caption segmentation, not guaranteed millisecond timing. Review silently for clear actions, listen without video for awkward language, then watch continuously with sound to confirm that “spoons” coincides with the spoon placement.

Make revision instructions specific. Replace “more natural” with “keep the three action lines and reduce the ending to one sentence without brand praise”. Replace “messy visuals” with “B02 already contains forks at its start; reshoot from an empty compartment with consistent counts”. For unclear captions, show only the current action word, avoid utensils and platform controls, and check placement in a full-screen phone preview.

Keep five deliverables: the fact sheet, final English narration, B01–B04 shot table, actual recording and reshoot or revision list. The handoff is ready when each sentence maps to an action, missing assets are marked and any difference from the target duration has a clear remedy.

When testing openings A and C, reuse confirmed facts and main footage where possible, and record published URLs and observation periods. For a save objective, compare save feedback rather than attributing an isolated high view count to the opening. These scripts are candidates; actual audience responses inform the next iteration.

Official reference and tutorial scope

OpenAI’s official Prompting guide recommends specifying the goal, context, output and boundaries, then reviewing and refining through follow-up messages. This tutorial applies that structure to product scripts.

Official documentation: https://learn.chatgpt.com/docs/prompting
The examples are filming plans and editorial demonstrations. Actual timing depends on recording and editing.