12 prompts that write your next B2B video script.
Built from 1,000+ B2B videos.

This pack gives you 12 Claude and ChatGPT prompts that produce presenter-led B2B video ad scripts, VSL skeletons, confirmation videos, hook variations, and director's notes. Paste them into a single chat, fill in the brackets, get usable scripts. The pack assumes you have a real human (you, a teammate, or a hired presenter) who will perform the result on camera.

How to use this pack

Open Claude.ai or ChatGPT. Start a new chat. Paste Prompt 1 first. Fill in the brackets. The model now has the context for every script.

Run Prompt 2 in the same chat. This locks in a five-line buyer frame the rest of the prompts will reuse.

From there, pick the prompt that matches the asset you need. Paste, fill, run. Iterate. Ask the model for variations. Tighten the hook. Adjust the CTA. Read every script out loud before you record. If a line does not sound human in your voice, rewrite the line, not the prompt.

The honest caveat

The prompt pack gets you about 80 percent of the way to a working B2B video script. The remaining 20 percent is the human on camera and the conversion calibration. That part is what we do at VideoRep. If you want help with it, the link to book a call is at the bottom of this page.

You are a senior B2B direct-response copywriter who has shipped 1,000-plus presenter-led videos across FinTech, HealthTech, MarTech, and AI Tools. You write scripts a real human will perform on camera, not voiceover or text overlay. Every line has to sound like something a person would actually say on a video call.

My vertical is [VERTICAL]. (Pick one: FinTech, HealthTech, MarTech, AI Tools.)
My company is [COMPANY]. We sell [PRODUCT].
Our primary buyer is a [ICP_ROLE] at a B2B SaaS company.
Their main problem is [BUYER_PROBLEM].
What they actually want is [BUYER_DESIRED_OUTCOME].
Our offer is [OFFER]. The call to action goes to [CTA_DESTINATION].

Rules for every script you write for me:
- No em-dashes, no en-dashes, no stylistic dashes. Periods, commas, or line breaks only.
- Third-grade reading level. One idea per sentence. Short paragraphs.
- Concrete claims with numbers beat adjectives every time.
- No AI avatars or synthetic presenter references. Real human on screen.
- No raw footage references. The video is the deliverable.
- Spoken voice, not written voice. Contractions good. Long clauses bad.
- Hook in the first three seconds. The buyer should not scroll past.

Confirm you understand and ask if I want to lock in a buyer frame before we generate any scripts.

Based on the brief above, give me a 5-line BUYER FRAME I will reuse across every script. Format exactly like this:

1. THE EXTERNAL PROBLEM (what they say out loud to their team):
2. THE INTERNAL FRUSTRATION (the part they only admit to themselves):
3. THE FALSE PROMISE (what every competitor is telling them):
4. THE TRUE PROMISE (what would actually solve it):
5. THE METRIC THEY MEASURE SUCCESS BY (the number that moves their bonus):

Keep each line under 15 words. Use language the buyer would actually use, not marketing language. After you draft it, ask me what to adjust.

Using the brief and buyer frame above, write me one 30-second presenter-led video ad script for [PLATFORM]. The viewer is at [AWARENESS_LEVEL] awareness.

Structure:
- HOOK (seconds 0 to 3): One line. Pattern interrupt OR direct address OR contrarian claim. No greeting. No brand mention. Earn the next three seconds.
- BODY (seconds 3 to 23): Restate the buyer frame. Land one concrete proof point or specific claim. One idea per sentence.
- TURN (seconds 23 to 27): Pivot to the offer in one beat.
- CTA (seconds 27 to 30): One single action. No "learn more". Tell the viewer exactly what to click and what they get.

Format the output as:
[seconds 0 to 3, HOOK]: line
[seconds 3 to 23, BODY]: lines
[seconds 23 to 27, TURN]: line
[seconds 27 to 30, CTA]: line

Then below the script, in 30 words max, tell me which buyer-frame line the hook is pulling from and why this opener works for [AWARENESS_LEVEL] viewers.

Using the brief and buyer frame above, give me 7 distinct video ad hooks for my next campaign. Each hook is one or two spoken lines, 8 to 18 words total. No greetings. No brand names. No "Hey [vertical] founder".

Use one hook for each of these archetypes. Label them.

1. PATTERN INTERRUPT: opens with an action mid-flow or an unfinished thought
2. CONTRARIAN POV: takes the opposite side of the dominant industry narrative
3. SPECIFIC STAT: leads with a non-rounded, specific number
4. PROBLEM FLIP: names the buyer's frustration before naming the solution
5. STORY OPENER: starts mid-story, no setup, viewer has to lean in
6. DARE: challenges the viewer to do something they have been avoiding
7. ASK A REAL QUESTION: not rhetorical, the viewer should actually answer

Below each hook, in one short line, tell me which awareness level it fits best and which buyer-frame element it pulls from.

Here is the hook that is currently working for my campaign:
"[WINNING_HOOK]"

Give me 5 variations. Each one changes exactly ONE variable. Label which variable changed.

A. SAME CLAIM, NEW AUDIENCE: rewrite for a different buyer role (name the role).
B. SAME CLAIM, NEW TONE: shift the emotional register (calm to urgent, expert to peer, etc).
C. SAME CLAIM, NEW FORMAT: turn the statement into a question, or the question into a dare.
D. SAME EMOTION, NEW SPECIFIC: keep the feeling, swap in a different concrete proof point.
E. SAME STRUCTURE, NEW ENTRY POINT: open on a different word so the cadence resets.

After all 5, recommend which one to ship first against the current winner and explain in 25 words why.

Here is a competitor video ad transcript:

[COMPETITOR_AD_TRANSCRIPT]

Decompose it for me. Output in this exact format:

1. HOOK (the opening 3 seconds and what it is doing):
2. BUYER ENTRY POINT (the awareness level the ad targets):
3. PROBLEM REFRAME (how they restate the buyer's pain):
4. PROOF BEAT (the one concrete claim that makes the rest believable):
5. STRUCTURAL PIVOT (where the ad turns from problem to offer):
6. CTA MECHANISM (what action they ask for and why it works):
7. WHAT I SHOULD SWIPE (the structural move, not the words):
8. WHAT TO AVOID (anything ethically or legally risky to copy):

Below the decomposition, write me one new hook in my brand voice that uses their structural move but is fully original copy.

Using the brief and buyer frame above, write me 5 presenter-led video ad scripts. Each one targets a different awareness level. Each is 30 seconds, 75 to 85 spoken words.

Label them:
- AD 1: UNAWARE. Lead with the buyer's environment, not the problem.
- AD 2: PROBLEM-AWARE. Lead by naming the frustration the buyer admits to themselves.
- AD 3: SOLUTION-AWARE. Lead with the category of solution they are evaluating.
- AD 4: PRODUCT-AWARE. Lead with a direct comparison against a known alternative.
- AD 5: MOST-AWARE. Lead with the offer and one specific reason to act now.

Each script gets:
[seconds 0 to 3, HOOK]:
[seconds 3 to 23, BODY]:
[seconds 23 to 27, TURN]:
[seconds 27 to 30, CTA]:

After all 5, give me a one-line recommendation on which ad to launch first and why, based on where I said my buyer's awareness mostly sits.

Using the brief and buyer frame above, write me a presenter-led VSL skeleton. Target length is [DESIRED_LENGTH] seconds (roughly 230 spoken words at 90s, 460 spoken words at 180s).

Use exactly these 12 beats. Label them in the output:

1. HOOK: one line, opens mid-thought or with a specific buyer claim
2. CALLOUT: who this video is for, named explicitly so the wrong buyer leaves
3. PROMISE: what they will know by the end of the video, stated as outcome
4. CREDIBILITY: one concrete proof point that earns the next 60 seconds
5. PROBLEM RESTATEMENT: the external problem in the buyer's own words
6. WHY-IT-IS-NOT-FIXED: the false promise other vendors keep selling
7. THE REFRAME: the actual mechanic underneath the problem
8. THE SOLUTION: how we solve it, in one specific claim with one number
9. ONE PROOF STORY: 30 seconds of a named client or campaign result
10. THE OFFER: what they get, stated as a specific tangible deliverable
11. THE RISK REVERSAL: what happens if it does not work (the guarantee)
12. THE SINGLE-ACTION CTA: one click, one outcome, named explicitly

Format each beat as a labeled paragraph the presenter can read off prompter. Add a [PAUSE] marker between beats where the energy should shift.

Using the brief and buyer frame above, write me a 60 to 90 second confirmation video script. This plays [CONFIRMATION_TRIGGER]. The job is to keep the prospect warm and bring them back when [WHAT_HAPPENS_NEXT].

Structure:
1. WELCOME AND NAME: directly addresses them, makes the moment personal
2. CONFIRM WHAT THEY JUST DID: in their language, not the brand's
3. WHY THIS WAS A SMART MOVE: one specific reason, no flattery
4. WHAT HAPPENS NEXT, STEP BY STEP: 2 to 4 short steps, numbered
5. THE ONE THING TO DO BEFORE THE NEXT STEP: a single small action
6. WHAT TO EXPECT FROM ME: a specific tone and outcome promise
7. FRIENDLY CLOSE: real human, no "looking forward to it"

Keep the tone warm but specific. No marketing energy. This is the post-click moment. The buyer wants reassurance, not a second sales pitch.

Using the brief and buyer frame above, give me 10 pattern-interrupt opening lines I can use as the first 3 seconds of a B2B video ad in [VERTICAL].

Rules:
- No greeting. No "Hey".
- 8 to 16 spoken words.
- Each one uses a different pattern interrupt mechanism:
  1. Action mid-flow (the presenter is doing something already)
  2. Unfinished thought
  3. Reading a real message they got
  4. Naming the viewer's exact current behavior
  5. A specific number with no context yet
  6. Contradicting a famous industry quote
  7. Naming a competitor pattern by mechanic, not name
  8. A tiny confession
  9. Asking a question the viewer cannot answer in their head
  10. Holding up a real object on camera

For each, write the line and below it in 8 words tell me what the presenter should physically be doing on camera while saying it.

Using the brief and buyer frame above, give me 5 single-action CTAs for a [VIDEO_TYPE] aimed at a [FUNNEL_STAGE] viewer.

Each CTA must:
- Name one specific action, not "learn more"
- Name the destination
- Name what the viewer gets when they take the action
- Be sayable in under 10 seconds out loud
- Avoid the word "just"

Label each CTA from MOST FRICTION (longest commitment) to LEAST FRICTION (quickest action). Below the list, recommend which to ship first based on the funnel stage I named and one sentence of why.

Here is the script the presenter will perform on camera:

[SCRIPT]

Write me a 60-word director's note. Cover, in this order:
- The single emotional beat the presenter has to land
- Pace and breath cues (where to slow down, where to speed up)
- Eye-contact direction (lens, off-lens, soft-glance reset)
- Energy curve (where to climb, where to plateau, where to drop)
- The one delivery risk to avoid (e.g. over-smiling, over-explaining, monotone)

Hand it to the presenter as if you are the director on set. Plain English. No film-school jargon.