TL;DR: Here are 45 AI prompts for conversion copywriting organized by channel — email subject lines, nurture and sales sequences, Google/Meta/LinkedIn ad variants, and landing page sections. Every prompt includes [bracketed variables] and the character constraints each channel actually enforces, so the output fits the format before you publish. Use them in ChatGPT, Claude, or Gemini.
What are the best AI prompts for conversion copywriting in 2026?
The best AI prompts for conversion copywriting are channel-specific templates that include the audience, the character limit, the angle rotation, and the output format — not generic "write me an email" or "write me an ad" requests. Conversion copy lives inside fixed containers: a Google RSA headline is 30 characters, a Meta primary text wraps at 125 characters above the fold on mobile, and an email subject line loses visibility after 50 characters on most clients. A prompt that ignores those containers produces copy that reads well in a chat window but fails in the actual channel.
Most marketers using AI for conversion copy fall into the same pattern: run a general prompt, get plausible-sounding output, spend 20 minutes reformatting it for the channel. This set is built the other way around. Every prompt includes the container specs the channel enforces, so the model produces ready-to-use copy the first time.
These 45 prompts cover the full conversion stack: 10 subject line prompts, 8 email sequence prompts, 17 ad copy prompts split across Google, Meta, and LinkedIn, and 10 landing page section prompts. They work in ChatGPT, Claude, or Gemini. For the broader marketing prompt set covering SEO, social, and research, our 50 ChatGPT prompts for marketers covers the full stack. If you want to A/B test these prompts against each other, our guide to prompt testing has the process. For marketers building a complete AI workflow, this set handles the conversion layer.
What do most conversion prompt lists get wrong?
Most listicles of "AI copywriting prompts" give you the general tasks — write a headline, write an email, write an ad — and stop there. What they miss is the channel container: the character limits, placement context, and audience signal that determine whether the output is actually usable.
A Google RSA headline is 30 characters. A Meta primary text wraps at 125 characters above the fold on mobile, where the majority of impressions happen. A LinkedIn Sponsored Content intro has a 150-character display before the "see more" break. These are not formatting footnotes — they are the context the model needs to produce copy that works without revision. Leave them out and you get copy that looks polished in the prompt window and fails the moment you paste it into the ad builder.
The second gap is sequence logic. Most lists give prompts for individual emails but nothing for the relationship between emails: what angle does email two take when email one went unanswered? The prompts in this set include that progression explicitly.
How do I write email subject lines with AI prompts?
Subject lines are the highest-leverage 50 characters in email marketing. These 10 prompts generate subject lines by type, segment, and behavioral trigger. Run them in batches and test the top three variants against each other — never publish the first output.
1. Subject line batch generator
Email type: [promotional / newsletter / transactional].
Segment: [describe — e.g., 7-day trial users who haven't used the core feature yet].
Offer or topic: [what the email is about].
Generate 20 subject line variants in 4 types: curiosity (5), urgency (5), benefit (5), question (5).
Constraint: 50 characters or fewer. No spam-trigger words ("free," "guarantee," "limited offer").
Rank the top 5 by predicted open rate. Explain each in one sentence.
Generate all 20 first, then use the ranked top five as your A/B test pool. Staying at 50 characters keeps subject lines intact on 90% of mobile clients.
2. Promotional campaign subject line
Campaign: [campaign name or offer — e.g., summer plan upgrade, ends [date]].
Target segment: [e.g., free users active in the last 30 days].
Angle options: scarcity, social proof, or curiosity gap.
Write 10 subject lines exploring all three angles.
Max 45 characters. Output: ranked list with one-sentence click-open rate rationale per subject line.
The angle options instruction forces three distinct emotional registers rather than ten variations of the same hook.
3. Welcome email subject line
Product: [product]. New user context: signed up but hasn't completed [key action].
Write 8 welcome email subject lines.
Mix: 3 action-prompting ("Start with X today"), 3 expectation-setting ("Here's what comes next"),
2 curiosity-gap ("The one thing most new users skip").
Max 50 characters.
4. Re-engagement subject line
Subscriber inactive for [90 / 180] days.
Last known action: [what they did before going quiet, if known].
Write 8 re-engagement subject lines. Avoid guilt. Avoid "we miss you."
Focus on: what changed, what they're missing now, or a low-friction re-entry offer.
Max 50 characters.
"We miss you" has become so overused it carries near-zero open lift. Asking the model to avoid it forces more specific angles.
5. Abandoned cart subject line set
Category: [product category — e.g., SaaS annual plan, e-commerce product].
Average cart value: [$X or plan name].
Write a 3-email subject line set with matching preview text:
E1 (1h): low-friction check-in. E2 (24h): address likely objection [price / complexity / timing].
E3 (72h): introduce offer or urgency if applicable.
Each: one subject line (50 chars max) + one preview text (35 chars max). No "you left something behind."
Preview text is what most prompt lists skip. It appears immediately after the subject in most clients and adds 3–5% to open rates when it adds new information rather than repeating the subject.
6. Post-purchase subject line
Product purchased: [product]. Days since purchase: [3 / 7 / 30].
Goal: [upsell / review request / onboarding nudge / referral ask].
Write 6 subject lines matching the goal. Tone: warm, not pushy.
Constraint: avoid the word "review" in the subject text of review-request emails.
Max 50 characters.
7. Newsletter subject line
Newsletter issue topic: [top 3 stories or themes].
Audience: [subscriber description].
Write 8 subject lines. Mix: 3 benefit-led, 3 curiosity, 2 list-format ("5 things this week").
Max 50 characters. Flag any variant that risks spam filters.
8. Cold outreach subject line
Prospect role: [job title]. Your specific value to them: [one sentence — not generic].
Write 8 cold outreach subject lines.
Rule: do not mention your company name or product name in the subject line.
Lead with their problem or a specific observation about their situation, not your solution.
Max 45 characters.
Removing the company name from cold subject lines often lifts open rates — it reads more like a peer referral and less like a mass campaign.
9. Subject line A/B challenger
Current subject line: [paste existing subject].
Current performance: open rate [X%], audience [description].
Hypothesis: what you think is underperforming (curiosity / personalization / length / angle).
Generate 5 challenger subject lines each isolating one change from your hypothesis.
For each: name the single variable changed and the predicted directional impact.
Changing one variable per test lets you isolate what actually moved the needle. Our guide to A/B testing your prompts covers the full methodology.
10. Upsell or expansion email subject line
Audience: existing [free / starter / mid-tier] plan users.
Upsell target: [plan or feature they don't have].
Write 6 subject lines that surface the gap without sounding like a sales email.
Approach: what they could accomplish with the upgrade, or what they're currently missing.
Max 50 characters.
What AI prompts write email sequences that convert?
A single email prompt gets you a draft. A sequence prompt set gets you a system — welcome series, nurture track, post-trial conversion flow — that runs while you focus on other things. These 8 prompts cover the most common sequences marketing teams build.
11. Five-email welcome sequence outline
Product: [product]. ICP: [describe — role, primary pain, goal].
Outline a 5-email welcome sequence:
E1 (D0): immediate value delivery. E2 (D1): one quick win. E3 (D3): deeper feature.
E4 (D5): social proof. E5 (D7): soft conversion or upgrade ask.
For each email: timing, subject line, body goal (30 words), one CTA.
Output as a planning table before writing any email body copy.
Build the outline first. It forces you to confirm the angle of each email before committing to body copy. The most common wasted effort in sequence-writing is drafting emails that overlap in intent.
12. Welcome email body (email one)
Context from sequence outline: [paste the E1 row from prompt 11].
ICP description: [paste].
Write the email body. Structure: one-sentence opener that delivers value immediately,
3 benefit bullets (outcome-led, not feature-led), one specific CTA.
Max 120 words. Tone: [paste voice brief or 4–5 adjectives].
13. Trial-to-paid conversion email
Trial length: [X] days. Trial ends in: [1 day].
Feature the user has not tried: [feature name and one-sentence description of what it does].
Write a 150-word conversion email. Open with what they did accomplish in the trial (specific action, not flattery).
Surface the untried feature. State the price and what they get. One CTA with plan name and price.
Do not open with "your trial is about to expire."
14. Three-email cart abandonment sequence
Cart value: [$X or tier name]. Most likely objection at this price point: [price / fit / timing].
Write a 3-email abandonment sequence:
E1 (1h): soft nudge, invite a question. E2 (24h): address the objection with a specific proof point.
E3 (72h): final call with discount or urgency if applicable.
Each: subject line + 100-word body + CTA.
15. Three-email re-engagement sequence
Inactive for [90] days. Last known action: [what they did].
Goal: identify still-interested from ready-to-churn, then act on each outcome differently.
Write 3 emails: E1 "what's new," E2 "here's what you're missing," E3 "last call or goodbye."
E3 should offer a graceful unsubscribe path. No guilt.
Each: subject line + 100-word body.
16. Seven-email B2B lead nurture sequence
Lead source: [inbound / content download / webinar registrant].
ICP: [role, company size, industry]. Core pain: [primary pain point].
Write a 7-email nurture sequence over 21 days. Each email: one angle only.
Angle rotation: problem education (2), solution framework (2), social proof (1), objection handling (1), CTA (1).
Output as a table: email number, send day, subject, angle, body summary (30 words), CTA.
17. Winback email for churned customers
Customer churned [30 / 90 / 180] days ago.
Reason for churning (if known): [price / not using it / found alternative / unknown].
Write a winback email. Open with what changed or improved since they left — not an apology.
Do not offer a discount as the opener; earn the offer with the value first.
Max 130 words. One CTA.
Opening with what changed is what separates winback emails that re-engage from the ones that read like support tickets.
18. Cold outreach three-email sequence
Prospect role: [title]. Company type: [describe].
Your specific value to this prospect: [1–2 sentences — not generic].
Write a 3-email cold sequence:
E1 (D0): 80 words, open with a prospect-specific observation, one CTA (reply or call link).
E2 (D3): 50 words, different angle — a result from a comparable customer.
E3 (D7): 40 words, close the loop and offer to leave them alone if now isn't right.
Tone: peer-to-peer.
How do I write Google, Meta, and LinkedIn ad copy with AI prompts?
Ad copy prompts fail when they ignore the container. Every network has different character limits, above-fold display rules, and audience context. These 17 prompts are structured by network so the output goes into the ad builder without reformatting.
19. Google RSA headline set
Product: [product]. Target keyword: [keyword].
Write 5 RSA headline groups (3 headlines each, 30 characters per headline, strict).
Include the target keyword in headline 1 of each group.
Angle rotation across groups: pain, benefit, social proof, price or offer, urgency.
For each group: 2 description variants (90 characters or fewer, strict).
20. Google Performance Max asset group
Product: [product]. Campaign goal: [awareness / conversion / app install].
Write one PMax asset group: 5 headlines (30 chars each), 5 long headlines (90 chars each),
4 descriptions (90 chars each), 2 short descriptions (60 chars each).
Angle each asset at a different benefit so Google's algorithm has distinct options to combine.
21. Google competitor comparison ad
Competitor: [competitor name]. Our specific differentiator: [verifiable, not "better quality"].
Write 3 search ads targeting "[competitor] alternative" and "[competitor] vs [us]" queries.
Each: 3 headlines (30 chars), 1 description (90 chars).
Do not rely on general superiority claims. One specific differentiator per ad.
22. Google brand keyword ad
Brand keyword: [our brand name / branded search variations].
Write 5 RSA variants capturing high-intent brand searches.
Each: 3 headlines + 1 description. Include one social proof element (rating, user count, or press mention)
in at least 2 variants. Route each to the most relevant landing page.
23. Feature launch search ad
New feature: [name and what it does in one sentence]. Target keyword: [keyword].
Write 3 ad variants announcing the feature.
Angle rotation: what the feature replaces (old pain), what it enables (new outcome), and speed or ease.
Each: 3 headlines (30 chars), 2 descriptions (90 chars).
24. Promotional urgency ad
Promotion: [offer — e.g., 30% off annual plan, ends [date]].
Target: searchers looking for [solution category].
Write 4 RSA variants with urgency. Include the end date in at least 2 headlines.
Constraints: 30 chars per headline, 90 per description. Use the specific date, not "hurry" or "limited time."
25. Single-keyword ad group (SKAG) copy
Keyword: "[exact keyword — include match type notation if relevant]".
Write 3 SKAG ads each mirroring the keyword directly in headline 1.
Each: headline 1 (keyword-matched, 30 chars), headline 2 (primary benefit, 30 chars),
headline 3 (CTA or social proof, 30 chars), 1 description (90 chars).
26. Meta TOF awareness ad
Product: [product]. Audience: [cold interest targeting — describe who they are].
Goal: awareness and page visit, not immediate conversion.
Write 3 Meta ad variants: primary text (125 chars above fold), headline (27 chars), description (27 chars).
Each variant: a different hook (problem-led, curiosity-led, social proof-led).
Add a one-line image brief for each variant.
27. Meta MOF retargeting ad
Audience: visited [specific page] but did not convert. Time in funnel: [1–3 / 7 / 30 days].
Write 3 retargeting variants addressing the most likely objection at this stage.
Primary text: 125 chars above fold. Headline: 27 chars.
Variant angles: price objection, social proof reinforcement, missed-feature reveal.
28. Meta BOF conversion ad
Audience: [added to cart / pricing page visit / trial signup with no conversion].
Write 3 bottom-of-funnel Meta ads. Open each by naming what the audience almost did.
Primary text: 125 chars above fold. Headline: 27 chars.
Include price, specific offer, and one trust signal per ad.
29. Meta carousel caption set
Product: [product]. Carousel purpose: [features / use cases / testimonials / before-after].
Write captions for a 6-card carousel.
Card 1: hook headline (12 words or fewer). Cards 2–5: one point each (8 words max). Card 6: CTA.
Each card: headline + sub-caption (25 words or fewer).
30. Meta UGC-style video ad script
Product: [product]. Customer use case: [how a real user benefits].
Write a 45-second UGC ad script.
Structure: problem (10s), discovery (5s), demo narration (20s), result (10s).
Tone: first-person, conversational, zero corporate language. No "game changer," "amazing," "life-changing."
Use specific outcomes, not adjectives.
31. Meta dynamic product ad text
Product category: [category]. Dynamic fields available: {product.name}, {product.price}, {product.brand}.
Write 5 primary text variants for dynamic product ads.
Each: 100 chars or fewer, references a dynamic field naturally, ends with an implicit or explicit CTA.
32. Meta video hook (first 3 seconds)
Video topic: [what the video shows]. Target: [audience].
Write 10 first-three-second hooks.
Each: one spoken line (12 words or fewer) + one-sentence visual description.
Label by hook type: question, shock stat, contrarian claim, problem-agitate, curiosity gap.
33. LinkedIn sponsored content (thought leadership)
Topic: [industry topic with a clear point of view]. Audience: [B2B role and company type].
Write LinkedIn sponsored content: intro text (150 chars above fold), body (200–280 chars), CTA button (25 chars).
Provide 3 variants. Tone: opinion-driven, not corporate. One provocative claim per variant.
34. LinkedIn lead gen form ad
Goal: download [asset] / register for [webinar] / request demo.
Audience: [job title, seniority level, industry].
Write: intro text (150 chars above fold), headline (70 chars), description (100 chars).
Headline should name the specific outcome the lead receives, not just the asset name.
35. LinkedIn retargeting ad
Audience: visited [page] or engaged with [post / company page] in the last [30] days.
Write 3 LinkedIn retargeting variants. Each: intro (150 chars), headline (70 chars).
Angle per variant: bridge from what they saw → what they gain by converting now.
What AI prompts build a landing page section by section?
Landing pages fail when copy is written all at once. Each section has a distinct persuasion job, and a prompt written for one job produces sharper output than a prompt trying to write the whole page. This is the conversion copy stack: hero → problem → solution → proof → objections → CTA, each with its own prompt.
36. Hero headline and subhead variants
Product: [product]. ICP: [describe in one sentence]. Core outcome: [what the customer achieves — specific].
Write 10 hero headline variants (12 words or fewer, outcome-led).
For the top 3: add a matching subhead (25 words or fewer) explaining the mechanism.
No buzzwords. No "the #1 tool for." Lead with the outcome, not the feature.
37. Value proposition block
ICP pain points: [list 3 specific pains in customer language, not marketing language].
Product: [product].
Write a 3-benefit block. For each: a headline (6 words or fewer) + two-line body (20 words).
Each benefit must lead with what the customer achieves, not what the feature does.
Flag any benefit where the customer outcome is unclear.
38. Problem-agitation section
Audience: [ICP]. Primary pain: [describe in their language — paste verbatim customer quotes if available].
Write a 100-word problem-agitation section for a landing page.
Structure: name the problem, describe what it costs (time, money, attention), heighten the consequence.
Tone: empathetic, not dramatic. Do not start with "Are you tired of." Do not use second-person accusatory framing.
39. Solution introduction paragraph
Problem just described: [paste from prompt 38 or summarize in one sentence].
Product: [product]. How it solves the problem: [specific mechanism, not vague].
Write a 75-word solution intro paragraph.
Structure: pivot from problem, name the solution, explain the mechanism in plain English, state the primary outcome.
No jargon. No "leveraging" or "powering."
40. Feature-to-benefit conversion block
Features list: [list 4–6 features].
ICP goal: [what they are trying to accomplish].
Rewrite each feature as a customer benefit: what the customer can now do, have, or avoid.
Format: feature → benefit (15 words max per benefit).
Flag any feature where the customer benefit is unclear or requires industry knowledge to understand.
41. Social proof block
Testimonials available: [paste 2–4 raw quotes].
For each: trim to 50–70 words covering pain, action, and outcome.
Add a one-sentence context line before each quote (role and use case if known; use case only if not).
Do not add metrics that are not in the original quote. Flag where a quote lacks a specific result.
42. Landing page FAQ section
Product: [product]. Top 3 objections from sales calls or support tickets: [list them].
Generate a 5-question FAQ for the landing page: the 3 objection questions plus 2 intent-matching questions.
Each answer: 40–60 words, conversational, direct. No "great question!" No "it depends."
43. Pricing section copy
Tiers: [name, price, and core differentiator per tier].
For each tier: a one-line positioning statement, 4 outcome-led feature bullets (what the customer achieves),
an "ideal for" line, and a CTA. Add a "most popular" marker where applicable.
Add one sentence per tier that preemptively answers "is this worth it at this price?"
44. Objection-handling section
Top 3 objections (verbatim from prospect conversations if possible): [list them].
Write an objection-handling section for the landing page.
For each objection: name it in the customer's own words, give a direct answer (40 words max),
add one proof element (data, quote, or feature detail). Do not use "We understand your concern."
45. CTA copy set
Funnel stage: [TOF / MOF / BOF]. Offer: [what happens when they click — trial, demo, download, purchase].
Write 8 CTA sets. Each: button text (4 words or fewer) + surrounding copy (15 words or fewer)
addressing the primary friction at this funnel stage.
Banned button copy: "Submit," "Click here," "Learn more."
How do I get more from these 45 prompts?
Three habits separate marketers who get usable first drafts from those who spend more time editing than the prompt saved them.
Fill every variable before running. A field that still reads [ICP] or [product] forces the model to invent an audience and product. It will produce the most generic version of your category. The bracketed fields are the signal, not decoration.
Add a voice brief to every customer-facing prompt. Paste three to five examples of your best-performing past copy before the prompt body. Conversion copy that sounds like your brand consistently outperforms copy that sounds like the average of the internet, and the examples are what anchor the model to your voice. Our brand voice Context guide covers how to build a reusable voice brief you stop rebuilding every session.
Chain prompts within a channel. Run prompt 36 (hero headline) before prompt 37 (value prop block) and pass the winning headline into the next prompt as context. The landing page reads as a coherent document rather than five independently generated sections.
| Habit skipped | What you get | Habit applied | What you get |
|---|---|---|---|
[ICP] left blank | Generic copy for no one | Specific ICP filled in | Copy written for one person |
| No voice brief | Average-internet cadence | 3–5 examples pasted in | Output you edit, not rewrite |
| Prompts run independently | Disjointed page sections | Prompts chained | Coherent page narrative |
| One output published | First-draft quality | Batch at temperature 1.0, pick best | 5 distinct options to test |
What mistakes do marketers make with conversion copy prompts?
Even with good templates, four errors quietly reduce output quality on every run.
- Leaving the channel container out. Writing "write me Google ad copy" without specifying 30-character headline limits means trimming every headline manually. Include the constraints in the prompt, not as a post-processing step.
- Using one prompt for a whole sequence. A single "write my 5-email welcome series" request forces the model to average the intent across all five emails. Individual prompts per email produce sharper, more distinct copy.
- Skipping the angle instruction. "Write 5 subject line variants" produces five versions of the same hook. Naming the angle types (curiosity, urgency, benefit, question) forces genuine variety — and variety is what A/B tests need to find a winner.
- Publishing the first output. Conversion copy is the genre where AI output most needs editing. The first draft gets you to 70%. The specific proof point, the line that sounds like a human made a genuine argument, the objection that matches what your sales team actually hears — that 30% is your job.
How Prompt Architects fits this workflow
All 45 prompts above work in any AI tool you already use. What Prompt Architects adds is the infrastructure that makes the set compound: a prompt library where you save each prompt with your standard variables pre-filled, Global Variables that store your ICP description, brand voice brief, and product one-liner once so they flow into any prompt automatically, and a Chrome extension that puts saved prompts one click away inside ChatGPT, Claude, or Gemini.
Half of our 2,170 customers had no prompt management system before signing up — not Notion, not a shared doc, nothing (our customer data, July 2026). For conversion-focused marketing teams, the prompt library is the first-order value: your subject line templates, your ad copy sets, and your landing page section prompts all in one place, reusable in 30 seconds instead of rebuilt from memory each campaign.
"The prompt library is genius — I save structured prompts by category and reuse them. Clean UI, no bloat. Just does the thing." — info.webefo, Verified AppSumo review
Prompt Architects is free to start, no credit card required.
Save the 10 prompts that apply to your highest-traffic channel first. Once you have run each one and saved the output format that worked, you have a reusable conversion copy system — not a set of prompts you rebuild from scratch next quarter.
Save these prompts to your library — free to start, works inside ChatGPT, Claude, and Gemini →