A bad abandon-cart email asks twice and bolts on a 5% discount. The average store recovers only 3-5% of carts that way, while top performers hit 10-14% (Mailmend, 2026) by finding the actual friction first — price doubt, shipping cost, trust, or decision fatigue — and matching the email to it. These 12 prompts force that step, and reserve the discount for when it really matters.
About 70.2% of carts are abandoned (Baymard Institute’s 2026 aggregate of 50 studies), so this is the single highest-leverage email flow most stores own. If shoppers bounce before they even add to cart, the upstream fix is usually the collection page — see AI Shopify collection descriptions for the “how to choose” framing that converts before they ever reach a product page.
TL;DR
- Send a 3-email sequence, not 1. A 3-email series recovers roughly 60-70% more revenue than a single send (Mailmend, 2026).
- Don’t lead with a discount. Email 1 should remove friction; save the offer for email 3, and only if the friction is genuinely price.
- Match the email to the abandonment reason. The top documented reasons are extra/shipping costs (39%) and forced account creation (24%) per Baymard.
- Paste any prompt below into ChatGPT (GPT-5.5), Claude (Sonnet 4.6), or Gemini 3.1 Pro, then feed in your real product, price, and shipping policy. Generic input produces generic email.
2026 recovery benchmarks (use these as targets)
| Metric | Typical store | Top performers | Source |
|---|---|---|---|
| Cart abandonment rate | ~70.2% | — | Baymard 2026 |
| Recovery (conversion) rate | 3-5% | 10-14% | Mailmend 2026 |
| Flow open rate | ~44.8% | up to 65% (Klaviyo top 10%) | Attribuly 2026 |
| Revenue per recipient | ~$3.65 | ~$28.89 | Klaviyo benchmark 2026 |
Treat open rate as soft signal only: Apple Mail Privacy Protection has inflated reported opens by roughly 10-15% since 2021, so judge success on clicks and recovered revenue.
Best for
- Shopify / WooCommerce / BigCommerce stores
- DTC brand email flows
- Subscription onboarding
- SaaS trial-to-paid
- B2B quote follow-up
1. 3-email sequence (1h / 24h / 72h)
The 1h / 24h / 72h cadence is a strong default for low-AOV, impulse-buy products. For considered or high-ticket items, push email 1 to 4-8 hours so it doesn’t feel pushy, and A/B test the first delay — that one variable moves recovery rate more than any copy tweak.
Design a 3-email abandon-cart sequence for [product / store]. Email 1 (1h): friction reminder + 1 link. Email 2 (24h): objection handler. Email 3 (72h): offer + scarcity (only if genuine). For each: subject, preheader, body under 120 words, CTA.2. Friction-finder pre-email
Before writing the email, help me think: for [product at $X with [Y] shipping], what are the top 3 reasons buyers abandon (price doubt, shipping cost, comparison shopping, account creation, security concern)? Pick the most likely and write the matched email.3. No-discount-first email
Write a 100-word email 1 that does NOT offer a discount. Goal: reduce friction (free returns, shipping policy, reviews) and remind them what they liked. Save the discount for email 3 if needed.4. Trust-signal email
Write a 110-word email that pivots to trust signals: real review, security badge, return policy, shipping speed. Use this when the friction is "is this legit?" rather than price.5. Objection-handler email
My most common abandonment reason is "[reason, e.g. not sure about size]". Write a 110-word email that proactively addresses this with: a size guide, an honest sizing note, and a 2-quote section from customer reviews.6. Comparison-shopper email
Write a 120-word email for the comparison-shopper. Format: 3-row comparison table vs the 2 likely competitors, with 1 honest concession plus 2 honest wins. End with a clear "if you need X, our product is built for that" line.7. SaaS trial-to-paid version
Write a 100-word email to a SaaS trial user who set up an account but did not subscribe. Focus on the 1 specific value they will lose on trial end, the 1 next step that locks in benefit, and a 1-question reply prompt.8. B2B quote follow-up version
Write a 130-word B2B follow-up to a buyer who requested a quote but did not respond. Format: name the date they requested, restate the proposal in 1 line, address the most likely internal blocker (budget cycle, security review), offer a 15-min call.9. Re-engagement after 30 days
Write a 90-word email for a cart abandoned 30+ days ago. Acknowledge time has passed, reference what is new (restock / new color / improved version), offer a soft return. Avoid guilt-trip language.10. Subject-line generator with preheader
Write 8 subject + preheader pairs for cart-abandon email 1. Variations: friendly nudge, curiosity gap, gentle FOMO, specific benefit, social proof. Subject under 50 chars, preheader under 90 chars. Mark which avoids spam triggers.11. Discount-only email (last resort)
Write a 90-word email offering [X]% off as the last-resort recovery. Frame it honestly (one-time, expires in [hours]). Avoid stacking with other promotions. Make the discount feel reasoned, not desperate.12. Post-purchase abandonment (cart cleared, not completed)
Write a 90-word email for a buyer who returned to clear their cart instead of completing. Hypothesis: they decided against. Ask 1 question — what would have changed their mind — and offer a no-pressure 10% off if they want to revisit.Which AI model to use
Any of the three big models writes solid email copy from these prompts; the difference is small and your input matters far more than the model.
- ChatGPT (GPT-5.5, $20 Plus): the default. Fast, good at tone variation, strong at the subject-line generator (prompt 10).
- Claude (Sonnet 4.6, $20 Pro): tends to write the most restrained, on-brand copy and follows the “no discount in email 1” constraint most reliably.
- Gemini 3.1 Pro ($19.99 Google AI Pro): cheapest API at $2/$12 per 1M tokens if you script bulk variant generation across a product catalog.
Always paste in your real return policy, shipping terms, and 2-3 actual review quotes. The model can’t recover a cart with placeholder text.
Common mistakes
- Leading with the same discount in email 1 (trains buyers to wait for it).
- Treating every abandoner the same — they leave for different reasons.
- Spam-trigger subject lines: ALL CAPS, “FREE”, “act now”, fake “RE:” prefixes, and excessive punctuation. In 2026, sender reputation and engagement weigh more than any single word, but stacked triggers still raise your spam score (GetResponse, 2026).
- No friction-specific copy — a generic “you left something behind” underperforms a reason-matched email.
- Sending more than 3 emails — diminishing returns plus rising opt-outs.
FAQ
How soon should the first email go out? Within about 1 hour for low-AOV impulse products; 4-8 hours for considered or high-ticket items so it doesn’t feel intrusive. Klaviyo’s own default is 2-4 hours; A/B test 1h vs 4h first, since the delay moves recovery more than the copy.
What recovery rate should I expect? Most stores recover 3-5% of abandoned carts. Well-tuned, friction-matched 3-email flows reach 10-14%, and the very best exceed 15-20% (Mailmend, 2026).
Does a discount really hurt? Only when it’s the opener. Leading every email 1 with a discount trains buyers to abandon on purpose. Keep email 1 discount-free and reserve the offer for email 3 — and only when the real friction is price.
Can I just send one email instead of three? You can, but you’ll leave money on the table. A 3-email series recovers roughly 60-70% more total revenue than a single send (Mailmend, 2026).
Which model writes the best copy? The gap is small. GPT-5.5 is the fastest all-rounder, Claude Sonnet 4.6 is the most disciplined on brand and constraints, and Gemini 3.1 Pro is cheapest for bulk catalog generation via API. Input quality matters more than the model.
Related
- Email marketing prompts
- Sales copy prompts
- Newsletter prompts
- CTA prompts
- Landing page section prompts
- Write Abandoned Cart Emails With AI
Tags: #Prompt #E-commerce #E-commerce