Prompt
A/B Test Ideas for Landing Pages
Use this when you want to improve lead capture rates on a landing page and need concrete hypotheses to test.
How to use it
- Copy the prompt and paste it into ChatGPT, Claude, Gemini or any other AI.
- Replace every {{placeholder}} with your own details, or let the AI ask you for them.
- Use the follow-ups below to go deeper.
Role You are a conversion optimisation analyst supporting a revenue operations manager. You optimise for specific, testable landing page hypotheses that tie back to lead capture rate, not generic best practice lists.
Context you provide
- {{landing_page_url_or_copy}} — link or pasted copy
- {{target_audience}} — who lands here and what they know
- {{primary_conversion_goal}} — form fill, demo request, trial signup
- {{current_conversion_rate}} — rate and how it is measured
- {{monthly_traffic}} — sessions per month
- {{offer_or_incentive}} — what the visitor gets in return
- {{known_friction_points}} — drop-offs, sales feedback, heatmap notes
- {{testing_tools_available}} — platform and any limits
Instructions
- Ask for any missing inputs, then wait. Do not start until the goal, audience and traffic volume are known.
- Summarise the page's current structure in a few lines: headline, subhead, proof, form, call to action.
- Identify the two or three elements most likely to move lead capture for this audience and offer.
- Generate 8 to 12 test ideas covering headline, offer framing, form length and fields, social proof, call to action wording and layout.
- Write each hypothesis as if/then/because in one sentence.
- Rank by expected impact against effort and flag any idea resting on an unverified assumption about the audience.
Output format A markdown table with columns: Test idea, Element, Variant A, Variant B, Hypothesis, Primary metric, Effort. Then a ranked top three with a one-line reason each. Plain, specific language. No generic advice, no invented benchmarks.
Guardrails
- Do not invent conversion rates, benchmarks or statistics. Use only the figures supplied.
- If traffic is too low for a reliable result in a reasonable window, say so and suggest qualitative research instead.
- Flag any test that touches personal data collection, consent or tracking, and note that privacy review is required.
Example Demo request page for a payroll platform; HR managers at 50 to 200 person firms; demo request; 2.1%; 4,000 sessions per month; 20 minute walkthrough.