Landing Page A/B Test Idea Generator
The most common way to waste an A/B test is pouring time into tiny changes like button color. Color rarely changes fate; headline, offer, and social proof do. The surface problem is 'what should I test'; the real problem underneath is 'which hypothesis moves conversion the most, and I should test that one first.' This prompt takes your existing page or email, extracts the hypotheses worth testing (headline, CTA, image, social proof), and ranks them by expected impact. It turns random testing habits into a prioritized testing culture.
You are a conversion rate optimization (CRO) expert. Your expertise: looking at a landing page or email and finding the test hypotheses that will move conversion the most, then prioritizing them. Your job is NOT to list every test you can think of; it's to produce hypotheses that are ranked by impact, individually actionable, and explained. I will give you: - [CURRENT PAGE/EMAIL CONTENT: headline, subheadline, CTA, social proof, what the image shows] - [PAGE GOAL: sale, signup, download, etc.] - [IF KNOWN: a known conversion problem or hunch] RETURN THE OUTPUT IN THIS ORDER, WITH THESE EXACT HEADERS: ═══ QUICK DIAGNOSIS ═══ The 1-2 most visible weaknesses undermining the page's conversion. ═══ TEST HYPOTHESES (RANKED BY IMPACT) ═══ For each hypothesis: what changes (headline/CTA/image/social proof), a concrete new version to try, why it might work, estimated impact (high/medium/low). ═══ TESTING DISCIPLINE NOTE ═══ One variable at a time, enough runtime, and locking in the winner before moving on. RULES: - Rank high-impact changes (headline, offer clarity, social proof) first; color and micro-copy go last. - Do NOT invent a made-up "this change increases conversion by X percent" figure; give estimated impact qualitatively as high/medium/low. - For every hypothesis, suggest real, ready-to-use alternative copy; don't give empty advice like "write a better headline." - When suggesting social proof, don't fabricate a fake review; say "add a real customer review here." - Write in natural, direct English.
How to use it
- Step 1: Hand over your current page or email copy as it stands: headline, subheadline, CTA, social proof, what the main image shows. The more complete your input, the sharper the hypotheses.
- Step 2: Trust the 'impact ranking' in the output. Run the high-impact tests first (headline, offer clarity); save low-impact tests (color, minor wording) for last.
- Step 3: Test one thing at a time. Change the headline and the CTA together and you won't know which one won. Lock in the winner, then move to the next test.
Example / tip
Input: Landing headline 'Quality leather wallets,' subheadline 'Handmade products,' CTA 'Shop Now,' no social proof. AI output (shortened): High-impact hypothesis 1, turn the headline into a concrete benefit, swap 'Quality' for 'A leather wallet that outlasts years of daily use and fits 3 cards plus cash'; the test rationale is a specific value over a vague adjective. Hypothesis 2, change the CTA from 'Shop Now' to 'Pick My Wallet,' making the action more specific. Hypothesis 3, add at least one real customer review to close the trust gap. Low impact: button color, saved for last.
Usage example
Copy the prompt as is, fill in the bracketed fields for your own business, and paste it into ChatGPT or Claude. The more concrete your input, the more useful the output. Take the first draft, ask for a one-sentence fix on whatever you don't like, then give it a final pass in your own voice.
When not to use it
Don't force it if you have no real input to work with. A prompt doesn't fill an empty field with data, it organizes and speeds up what you already know. Never publish the output unchecked; AI sometimes produces a claim that sounds right but is wrong or made up. Always verify anything that needs proof yourself: numbers, prices, guarantees.
Output quality checklist
- Is the output specific to your input, or generic filler that fits any business? If it's generic, run it again with more concrete input.
- Does every number, ratio, or claim have something backing it? Delete the unbacked ones or soften them honestly.
- Does it sound like you? Simplify the cliches and inflated lines in your own voice.
- Does it give one clear next step? If the reader knows what to do, the output did its job.
